ABOUT THIS LEARNING OBJECT
AIS4YW is an Erasmus+ KA210-YOU small-scale partnership that strengthens the quality and innovation of youth work through hands-on, ethical use of generative AI in non-formal education. The 20-month project runs from 4 March 2024 to 3 November 2025. It is led by Associazione Arcipelago APS (Italy) with Fundación Esplai, Ciudadanía Comprometida (Spain) as partner. Priorities include improving youth-work quality and promoting inclusion and diversity. AIS4YW responds to the urgent need for youth workers (YWs) to understand how generative AI works, what opportunities it offers, which risks it entails, and how to deploy it critically, ethically, and safely with young and very young people.
The project is built as a community of practice between Italy and Spain to share methods and resources for professional, conscious AI use in non-formal learning. Direct participants are 16 youth workers (Italy & Spain) engaged through learning by doing, peer learning and “in-situation” practice. The content you will see below is the result of the exploration and practice carried out over the past months, which we hope will be useful to your Youth Work daily practice.
Table of contets
Chapter 1 - Introduction
Chapter 2 - Impact of AI in non-formal education
Chapter 3 - GymComp: Generative AI for Non-Formal Education
Chapter 4 - Ethics of AI Systems
Chapter 5 - Privacy & IA
Chapter 6 - Netiquette and Artificial Intelligence
This table of contents should allow you to easily navigate through the contents of this Learning Object. Browse through the chapters and find interactions with the content through images and icons.
Chapter 7 - Tools and practical examples
References
Chapter 1 - Introduction
AIS4YW is an Erasmus+ KA210-YOU small-scale partnership that strengthens the quality and innovation of youth work through hands-on, ethical use of generative AI in non-formal education. The 20-month project runs from 4 March 2024 to 3 November 2025. It is led by Associazione Arcipelago APS (Italy) with Fundación Esplai, Ciudadanía Comprometida (Spain) as partner. Priorities include improving youth-work quality and promoting inclusion and diversity. AIS4YW responds to the urgent need for youth workers (YWs) to understand how generative AI works, what opportunities it offers, which risks it entails, and how to deploy it critically, ethically, and safely with young and very young people. The project is built as a community of practice between Italy and Spain to share methods and resources for professional, conscious AI use in non-formal learning. Direct participants are 16 youth workers (Italy & Spain) engaged through learning by doing, peer learning and “in-situation” practice. Benefits extend to partner staff, local networks and broader European communities through the LO and dissemination events.
Objectives
Key content/discussion
Conclusion
Chapter 2 - Impact of AI in non-formal education
The impact of Artificial Intelligence (AI) on youth work and education is no longer a theoretical debate—it is a lived, dynamic reality. Building on the foundational work of the AI4YouW project, which piloted and validated open educational resources (OERs) with over 100 youth workers across Europe, the current AI4YW initiative expands that vision by actively engaging professionals in reflective, practical, and ethically grounded experimentation with generative AI tools. Through co-designed workshops held in Spain and Italy, educators explored how AI can reshape educational design, content creation, youth engagement, and professional development. These experiences were informed by global policy frameworks such as UNESCO’s 2023 guidance on generative AI in education, and conceptually anchored in the European Digital Competence Framework for Citizens (DigComp 2.1) and the LIFEComp framework for personal, social and learning-to-learn competences. Together, these frameworks provided the foundation for the GymComp methodology, a gamified and participatory process used to help youth workers reflect on and develop the competences needed for the responsible and effective use of AI. This chapter draws on diverse sources—workshop results, impact matrices, and pilot evaluations—to offer a grounded perspective on the opportunities, challenges, and transformative potential of AI. Rather than endorsing technological determinism, it advocates for a critically informed, inclusive, and human-centered approach to AI integration in youth work.
Key content/discussion
Conclusion
Chapter 2 - Practical examples
Example 3
Example 1
Example 2
Gamma – enables dynamic, visual presentation
Chatgpt Learning Experience Designer
Brisk Teaching – supports active learning through integrated micro-learning activities
Example 4
Example 5
Whimsical – facilitates the development of concept maps during group work
Mizou – supports the creation of interactive mock-ups for AI–persona interaction scenarios
Chapter 3 - GymComp: Generative AI for Non-Formal Education
This chapter introduces Gym Comp—a practical, hands-on approach to strengthening young people’s digital and relational capacities through short, repeatable training “circuits” that use generative AI in non-formal learning spaces. We translate key strands of the European competence frameworks—DigComp 2.2 (1.1–1.3 information and data literacy, 2.3 online participation, 3.1 digital content creation, 4.2 privacy and data protection) together with LifeComp’s communication, collaboration, empathy, and self-regulation—into activities that youth workers can run with minimal prep and maximum impact. Each circuit blends critical search and verification, prompt-driven creativity, and privacy-by-design routines aligned with GDPR, while openly addressing GenAI’s risks (bias, hallucination, copyright, safety) and showing how to mitigate them in everyday practice. The work presented here has been co-designed and piloted with partners in Italy and Spain. In Italy, the programme is curated by Associazione Arcipelago APS in collaboration with Lascò, Professor Raffaele Mele and Mattia Anicito, who together have adapted Gym Comp to community centres, youth clubs, libraries, and after-school contexts. In Spain, the roll-out and localisation are led by Fundación Esplai drawing on its long experience in digital inclusion and youth empowerment. This cross-country collaboration keeps the method grounded in real settings and diverse learner needs, while ensuring that the examples, scenarios, and protocols remain ethically sound and culturally relevant.
Key content/discussion
10
Reflexion
Conclusion
Chapter 3 - Appendices
Digicomp 2.3. - Participainting citizenship through digital technologies
DigiComp - Information and data literacy
Introdcution, definition and competence construct
LifeComp
DigiComp 4.2. - personal data proteccion and privacy
DigiComp 3.1. - Digital Content Creation
Chapter 4 - Ethics of AI Systems
The ethical dimension is one of the most relevant aspects of the impact of information technologies on social organization. The ethical aspect was little considered in the past, as people believed in the paradigm of 'self-regulation' of IT products, services, and activities, promoted for decades by the major tech giants. The scenario began to change thanks to two key factors: - Social awareness of the unforeseen negative effects of digital technology (e.g., information manipulation, loss of privacy, inequalities in access). - The global expansion of AI and, above all, of generative AI, which made evident the risks of a technology capable of making decisions or generating content with social, legal, and economic impacts. This changing scenario has made it necessary to raise ethical questions about the use of AI, even in fields outside of technology where AI has expanded and become essential in daily life, such as education (formal and non-formal).
Key content/discussion
Reflexion
Conclusion
Chapter 4 - Appendices
Ted Talks 1
Guiding Questions for Youth Workers on the Ethical Use of AI
Emerging competences in ethical use of AI and data for YWs
Ted Talks 2
Chapter 5 - Privacy and AI
The introduction and growing spread of artificial intelligence systems within today’s economic and social context have produced a significant change in the ways personal data are collected, processed, and used, raising regulatory issues of considerable complexity. The European legislator has chosen to respond to these challenges with a multi-level regulatory model, based on the interaction between the General Data Protection Regulation (GDPR) and the new Artificial Intelligence Regulation (AI Act), recently approved by the European Parliament as part of a broader EU digital strategy. Privacy protection exists only if it is guaranteed by laws, regulations, and directives that establish which data can be collected and in what ways.
Key content/discussion
Reflections
Conclusion
Chapter 6 - Netiquette and Artificial Intelligence
Netiquette is an English word that combines the English term network and the French word etiquette (good manners). It is a set of informal rules that regulate good user behavior on the web, especially in interacting with other users through resources such as newsgroups, mailing lists, forums, blogs, social networks, or email in general. Following the guidelines for acceptable behavior makes the internet a more pleasant place for all users. The way we interact with AI – and the way AI interacts with us – requires a new set of good manners
Key content/discussion
Conclusion
Chapter 7 - Tools and Practical Examples
This chapter presents a curated selection of tools, practice sheets, and tutorials that stem from the collaborative work of the AIS4YW project partnership. Developed in the context of non-formal education, the materials aim to support youth educators, trainers, and facilitators in integrating Artificial Intelligence (AI) into their educational practices in a thoughtful, inclusive, and ethically grounded way. The resources collected here reflect the core themes explored throughout the project, including the impact of AI on youth work, the development of digital and transversal competences (inspired by frameworks such as DigComp and LIFEComp), and the ethical and responsible use of digital technologies. The tools are aligned with these dimensions and were either co-created or selected by the project partners based on real needs identified during workshops, collaborative sessions, and local experimentation. Rather than offering an exhaustive inventory, this chapter highlights practical, ready-to-use resources — including AI-based tools, educational platforms, and examples of real-life applications — that can inspire educators in non-formal settings. The selected materials are adaptable to different contexts and learning environments and are designed to foster critical thinking, creativity, and learner empowerment. The chapter is structured in two sections: an overview of different toolkits and a collection of practice sheets on how to explore the use of AI with youth work. These sections aim to provide both conceptual clarity and hands-on guidance to those seeking to integrate AI in educational experiences beyond a non-formal context.
Chapter 7 - Toolkit overview
AI Tools for Collaboration and Communication
Educational Platforms Integrating AI
AI Tools for content creation
AI Tools for Ethical Reflection and Critical Thinking
AI Tools for Assessment and Feedback
Chapter 7 - Practice sheets
My Digital Competence Self-Assessment (Based on GymComp)
Ethical Dilemmas and AI
Exploring AI in Daily Life
Designing an AI-Enhanced Workshop
AI Tools in Practice: A Critical Test
References
Chapter 2
- UNESCO. (2023). Guidance for generative AI in education and research. UNESCO. unesco.org+1
- Carretero, S., Vuorikari, R., & Punie, Y. (2017). DigComp 2.1: The Digital Competence Framework for Citizens with eight proficiency levels and examples of use (JRC Technical Report). European Commission. IDEAS/RePEc
- European Commission / Joint Research Centre. (n.d.). LifeComp: The European framework for personal, social and learning to learn key competence. JRC / European Commission. joint-research-centre.ec.europa.eu
- Kechagias, K. (2025). Artificial intelligence competence needs for youth workers. Journal of Non-formal Education. journals.team4excellence.ro
- Ghimire, A., & Edwards, J. (2024). From Guidelines to Governance: A Study of AI Policies in Education. arXiv preprint. arxiv.org
- Perkins, M., Furze, L., Roe, J., & MacVaugh, J. (2023). The AI Assessment Scale (AIAS): A Framework for Ethical Integration of Generative AI in Educational Assessment. arXiv preprint. arxiv.org
- Bura, C., & Myakala, P. K. (2024). Advancing Transformative Education: Generative AI as a Catalyst for Equity and Innovation. arXiv preprint. arxiv.org
Chapter 3
- European Commission, Joint Research Centre. (2022). *DigComp 2.2: The Digital Competence Framework for Citizens (JRC128415).
- Sala, A., Punie, Y., Gualtieri, M., et al. (2020).LifeComp: The European Framework for Personal, Social and Learning to Learn.
- High-Level Expert Group on AI. (2019). *Ethics Guidelines for Trustworthy AI. European Commission.
- European Commission. (2020). Assessment List for Trustworthy Artificial Intelligence (ALTAI) – Self-assessment. Publications Office / Digital Strategy. ([Publications Office of the EU][5], [Strategia Digitale Europea][6])
- European Commission. (2022). Ethical guidelines on the use of AI and data in teaching and learning for educators.
- European Union. (2016). Regulation (EU) 2016/679 (General Data Protection Regulation).
- European Union. (2002/2009). Directive 2002/58/EC (ePrivacy Directive)
- European Union. (2022). Regulation (EU) 2022/868 (Data Governance Act). Official Journal; see also Commission explainer.
References
Chapter 4
- Ethical guidelines on the use of artificial intelligence (AI) and data in teaching and learning for educators (2022)
- GUÍA BÁSICA DE LA IA, de los textos, Aguilar, I.; Alepuz, V.; Alfaro, J.; Bañón, J.J.; Botti, V.; Despujol, I.; Giménez, J.; Linares, J.; Linares, J.M.; Majadas, V.; Martínez, J.; Monsoriu, M.; Montesa, E.; Morillas, C.; Muñoz, J.M.; Ortega, J.; Ortuño, A.; Peñarrubia, J.P.; Plasencia, A.; Rieta, J.; Sales, M.; Segarra, R. (2024).
- Edita Smart Digital.Data Strategy: A Blueprint for an EU data modelJRC (2022).
- DigComp 2.2: The European Digital Competence Framework for CitizensOECD (2021).
- Recommendation of the Council on Artificial IntelligenceUNESCO (2021).
- Recommendation on the Ethics of Artificial IntelligenceUNESCO (2019).
- Artificial Intelligence in Education: Challenges and Opportunities for Sustainable DevelopmentUNICEF (2021).
Chapter 5
- "A. Augostini, G. Minucci, J. Giussani, G. Lombardi, The Relationship between Artificial Intelligence and Personal Data Protection, Lexia, 20/11/2024,
- An Introduction to the Code of Practice for General-Purpose AI
- EU AI Act: first regulation on artificial intelligence
References
Chapter 6
- The art of AI etiquette – WeAreTechWomen (2025)
- Microsoft Enterprise AI Services Code of Conduct (2025)
- 8 AI etiquette & content tips – HRMorning (2023)
Get in touch with the project managers in each country: Italy -Arcipelago info@associazionearcipelago.com Spain - Fundación Esplai gporres@fundacionesplai.org
5) Ethics, rights and trustworthy-by-design
The materials adopt the EU’s ALTAI checklist to keep creation and participation “trustworthy”: respect fundamental rights (including minors), maintain human oversight, ensure technical robustness and safety, apply privacy-by-design/default, guarantee traceability and transparency, promote diversity/non-discrimination and consider social-environmental impacts. We convert these into checklists embedded in the circuits (e.g., “Who is affected?”, “Where are the data from?”, “Where is the human in the loop?”). Partners repeatedly note that learners must see how systems work—contrasting black-box with interpretable models, experimenting with simple training tools—to avoid “magical thinking” about AI.
DigiComp 4.2. - Personal data protection and privacy
Why AI ethics is crucial
AI is not just powerful software: it is a system capable of influencing decisions that once belonged exclusively to human beings. This implies direct consequences on fundamental rights such as: - Privacy, since AI processes enormous amounts of personal data; - Equal treatment, because uncontrolled algorithms can replicate or amplify social biases; - Safety, as errors or manipulations in AI systems can generate material or moral harm on a large scale. In recent years, it has become clear that 'relying on the common sense of technology creators' is not enough. What is needed are: - Clear regulatory frameworks, defining limits and responsibilities; - Shared ethical standards, guiding design, development, and implementation; - Transparency and audit mechanisms, enabling verification by independent entities.
CHAPTER 1
Conclusion
AIS4YWs turns big frameworks into small, durable habits. With GymComp, youth workers practise verify → create, co-govern participation, and privacy by default until these moves become routine; with the Workshops, they package what works into a multilingual Learning Object other practitioners can adopt the next day. Together, these strands link DigComp 2.2 and LifeComp to everyday facilitation, keeping ethics, safety and inclusion inside the work rather than bolted on. This approach squarely serves the Erasmus+ priorities—quality and innovation in youth work and inclusion & diversity—by building a shared, cross-border method rather than a tool list. Roles are clear (Arcipelago coordinates GymComp and dissemination; Esplai leads monitoring, impact, and the LO) and the timeline embeds learning, evidence, and dissemination events that carry results beyond the cohort.
The project’s value also travels further: sessions and assets plug into existing youth programmes, civic education, libraries and community centres, while partner networks in Italy and Spain amplify reach and sustain practice after funding. In short, AIS4YWs offers a practical bridge from European policy to street-level education: repeatable circuits, lightweight evidence, and open resources that help young people create, verify, participate—and protect themselves and others—using AI responsibly If the sector holds to three commitments—verify then create, privacy by default, and co-governed participation—youth workers don’t need to be AI engineers to lead meaningful learning. They need a humane method, shared assets, and peers across Europe. AIS4YWs contributes all three.
Transparency, fairness, and reliability
Reliability
- An AI system must function safely and consistently over time.
- Risk mitigation plans, regular updates, and independent audits must be provided.
In addition, further key aspects are necessary to ensure AI is reliable and safe:
- Respect for human autonomy – All systems must allow for human oversight, especially in decisions affecting minors and young people.
- Social and environmental well-being – Avoiding social harm, promoting cohesion and sustainability.
- Data privacy and governance – Protection of personal data, quality and integrity of data, controlled access.
- Accountability – Clear responsibilities must exist for monitoring, negative impacts, and appeals.
Three pillars are essential to ensure an AI system is ethical: Transparency
- Users and authorities must know how and why a system makes certain decisions.
- Opaque systems (black boxes) must be accompanied by explainable AI (XAI) tools.
- This aspect is fundamental for trust and the exercise of rights.
Fairness
- Algorithms must not generate discrimination or reinforce bias.
- It is crucial to test models on diverse datasets and to constantly update evaluation criteria.
CHAPTER 5
Reflections
The concrete implementation of the provisions established by the GDPR and the AI Act raises significant technical and operational challenges, requiring the development of innovative solutions and the adoption of integrated approaches to regulatory compliance. Ensuring both compliance with data protection principles and adherence to the specific requirements established for artificial intelligence systems, in fact, entails designing technological architectures and organizational frameworks capable of jointly meeting the needs imposed by both sets of regulations.
Chapter 1
Key content/discussion
GymComp sits at the heart of AIS4YWs—a competence “gym” that turns the EU frameworks into short, repeatable, phone-friendly routines youth workers can run in real spaces with real young people. Rather than being a course “about AI models,” it’s a way to consistently do three things well: verify before you create, participate responsibly, and protect people and data. The design starts from concrete youth-work problems (myths like “AI knows everything,” privacy fatigue, licensing limits) and only then maps back to the frameworks, so competences don’t remain labels but become habits under constraint. In practice we work across DigComp 2.2 areas 1.1–1.3, 2.3, 3.1, 4.2 and weave in LifeComp’s social–emotional strands, so sessions build both technical and relational capacity. Each GymComp “circuit” is a brisk lab with clear timeboxes, rotating roles (scribe, checker, privacy lead), and a visible exit ticket tied to competence descriptors. In Critical search & verification, groups compare a web search, a conversational model, and a citations-first engine, tracing every claim back to sources. Civic participation online tackles real local issues via democratic platforms (Decidim-style) and co-writes fair-use rules to avoid popularity contests. Create with care prototypes a mini-campaign and addresses authorship, derivative works, and the ethics of generated media. Safety by default is a rotating-station lab on cookies/consent, messaging-app settings, password managers, and metadata. A final strand, Prompting for truthfulness, builds guardrails against hallucination, bias, copyright exposure, and data leakage (roles/audiences/formats, “require citations,” retrieval for grounding). These flows were piloted in Italy and Spain and iterated to fit free vs. licensed tools and different learning contexts. Ethics and rights are kept inside the practice, not bolted on. Trainers collapse the EU ALTAI checklist into a few prompts—Who is affected? Where are the data from? Where is the human in the loop?—so dignity, transparency, oversight, robustness, and non-discrimination guide creation and participation in real time. Privacy is taught as a habit: unique passwords and managers, reading consent flows, understanding differences among messaging apps (E2E defaults, backups, metadata), and configuring settings on the spot. A simple “three promises” mnemonic helps: collect less, share consciously, retain briefly—illustrated with everyday cases like cookie choices and algorithmic recommendations.
The project scaffolds this method with a clear structure, roles, and outputs. Partners run six GymComp sessions × 3 hours (18 hours total) for 16 youth workers from Italy and Spain, using learning-by-doing, peer learning and “in-situation” practice. The same cohort meets for four online Workshops (2 hours each) to co-create a Learning Object (LO)—a multilingual (EN/IT/ES) bundle of materials, methodologies, and good practices for safe, ethical AI use in non-formal education. Arcipelago leads GymComp delivery; Fundación Esplai coordinates the Workshops and assembles the LO for open access via partner websites. Workshops are deliberately production-oriented: participants consolidate what worked in GymComp, document the steps, and turn procedures into reusable resources for peers. The emphasis is on transfer—from a facilitation routine tested in a youth club to a clearly documented method other youth workers can adopt the next day—rather than on showcasing specific apps. This keeps the project tool-agnostic: the process (plan → co-create → verify → attribute) matters more than brand names, which also helps maintain inclusion when free tiers shrink or features move behind paywalls. Across sessions, learning remains visible but light-touch: quick polls to set the pace, exit tickets aligned to DigComp descriptors, and short self-assessments to surface growth. For 3.1 tasks, rubrics look at intent clarity, licensing/attribution, audience fit, and transparency about AI assistance; for 4.2, evidence includes changed settings, stronger password hygiene, and a short “privacy story” (what changed and why). The result is evidence without bureaucracy—and behaviours that stick once devices are back in pockets. In the end, GymComp plus Workshops form a virtuous loop: situated practice → lightweight evidence → a shared, portable LO. This loop is what lets AIS4YWs keep its promises—quality youth work, inclusion and diversity, and a critical/ethical/safe use of AI—by turning European frameworks into everyday routines that are verifiable, transparent, and inclusive.
CHAPTER 3
Reflexions
Gym Comp works because it treats EU frameworks as means, not ends. The materials start from real youth-work frictions—privacy fatigue, tool access limits, “AI knows everything” myths—and translate DigComp/LifeComp into repeatable routines that learners can actually do under time pressure. That is the decisive shift: competences are not lists on a wall but behaviours embedded in short circuits—check the source, name the risk, set the control, reflect as a team. This framing mirrors LifeComp’s awareness–understanding–action arc and makes the “relational layer” visible rather than ornamental. A second takeaway is the centrality of safety-by-default. When privacy and security appear only as an add-on, they lose to convenience; when they are baked into every task (cookies, consent, passwords, messaging settings), learners start to narrate their own privacy stories and make concrete changes (e.g., enabling MFA, rotating passwords, choosing E2E defaults). The Italian privacy deck helps by turning abstract rules (lawful basis, purpose limitation, minimisation, storage limits) into short, memorizable prompts that participants can reuse across platforms. In practice, teaching the why of GDPR without the how of settings and habits is ineffective; Gym Comp keeps them coupled.. The sessions also surface an equity issue often ignored in AI trainings: access and licensing. Partners document how free tiers shrink and upload limits change, pushing learners toward paid plans. That matters in youth work, where the promise of “AI for everyone” collapses if activities depend on gated features. The honest response is not to lament but to design tool-agnostic exercises (verify, attribute, protect) and plan for institutional accounts or open alternatives when a licensed tool is pedagogically necessary. This keeps inclusion non-negotiable, not aspirational.
On the ethics and trust front, the ALTAI lens proves practical when collapsed into a few recurring questions: Who is affected? Where are the data from? Where is the human in the loop? What can go wrong, and how will we notice? In workshops, these prompts helped move from “AI is cool/scary” to auditable practice—traceable outputs, explicit attribution, and an agreed human review step. Rather than policing language, the checklists scaffold judgement and allow youth workers to justify choices to peers, funders, and participants.Still, AI risk literacy must be taught as technique and mindset. Slides on “modern AI hype” make a useful distinction: models can dazzle on unstructured data and still hallucinate confidently; learners need both prompting discipline and external grounding to resist that pull. In our view, the most robust habit taught here is “verify first, then create”: start from sources, constrain the model with them, and cite. This counters magical thinking without dampening creativity. The civic participation strand is where facilitation is most tested. Spanish discussions show how participatory platforms can drift into popularity contests and rivalry without guardrails. That is not a reason to abandon them; it is an argument for explicit norms, transparent data use, and bias-aware moderation designed with young people, not for them. Gym Comp’s choice to prototype participation rules inside the session is the right move: it treats youth as co-governors of the digital spaces they use.
CHAPTER 6
Conclusions
AI netiquette is not just a set of rules, but a way to build a harmonious relationship between humans and machines.
- Transparency builds trust.
- Respect for privacy ensures security.
- Fairness promotes equality.
- Responsibility and continuous improvement keep AI effective and ethical.
- The human touch ensures empathy and understanding.
As AI grows, these principles will guide us to use it responsibly and ethically. Welcoming AI with awareness means making it a partner, not just a tool. It is a way to build a future where AI benefits everyone.
2. Challenges: bias, inequality, and overdependence
Despite its promise, AI also presents substantial challenges that were discussed at length during workshops: Participants identified the risk of overreliance on AI outputs, especially by youth with limited critical thinking skills. Several youth workers shared concerns about young people treating AI responses as facts without questioning their source or limitations. Digital inequality emerged as a critical barrier: while some youth workers felt empowered by AI tools, others lacked the foundational digital skills to engage meaningfully with them. This divide risks widening existing social and educational inequalities unless proactive scaffolding is provided. The ethical risks of data misuse and lack of transparency in algorithmic decision-making were also flagged, especially when AI tools are used in sensitive contexts involving minors or marginalized groups. These concerns were mirrored in the AI4YouW pilot, where participants called for stronger guidance on privacy, clearer ethical use protocols, and more culturally adapted content for diverse learning communities
3) What learners do: core Gym Comp circuits
Circuit A — Critical search & verification (DigComp 1.1–1.3, 2.3; LifeComp: critical thinking). Learners compare answers from Google, a conversational model, and a citations-first engine, then trace claims back to sources. Partners found that tools which display sources transparently help demystify how results are produced and let learners judge credibility. Circuit B — Civic participation online (DigComp 2.3). Starting from local issues, groups explore democratic platforms (e.g., Decidim-style processes), reflect on pitfalls such as “popularity contests” in neighbourhood voting, and draft contribution guidelines that minimise bias and rivalry. Circuit C — Create with care (DigComp 3.1; LifeComp: collaboration & communication). Teams prototype a short campaign (poster, post, 30-sec video). Trainers introduce authorship, derivative works and the ethics of generated media; learners practise spotting AI-made vs human-made images and discuss fairness in contests where AI is used.
Circuit D — Safety by default (DigComp 4.2; LifeComp: self-regulation). A rotating station lab: manage cookies and consent banners; tune privacy on messaging apps; test password strength and managers; identify platform metadata and their implications. Slides summarise GDPR principles, ePrivacy scope, rights of the data subject, and pragmatic routines learners can apply immediately. Circuit E — Prompting for truthfulness. Learners see typical failure modes (hallucination, copyright, privacy leakage, bias, and cost trade-offs), then practise guardrails: specify role, audience and format; require citations; and use a retrieval step to ground outputs in provided materials. Each circuit includes timeboxes, roles (scribe, checker, privacy lead), and a visible exit ticket aligned with the competence descriptors. Italian and Spanish cohorts tested these flows in youth clubs and workshops, iterating based on learner feedback and the practical limits of free vs licensed tools.
Chapter 1
Key content/discussion
GymComp sits at the heart of AIS4YWs—a competence “gym” that turns the EU frameworks into short, repeatable, phone-friendly routines youth workers can run in real spaces with real young people. Rather than being a course “about AI models,” it’s a way to consistently do three things well: verify before you create, participate responsibly, and protect people and data. The design starts from concrete youth-work problems (myths like “AI knows everything,” privacy fatigue, licensing limits) and only then maps back to the frameworks, so competences don’t remain labels but become habits under constraint. In practice we work across DigComp 2.2 areas 1.1–1.3, 2.3, 3.1, 4.2 and weave in LifeComp’s social–emotional strands, so sessions build both technical and relational capacity. Each GymComp “circuit” is a brisk lab with clear timeboxes, rotating roles (scribe, checker, privacy lead), and a visible exit ticket tied to competence descriptors. In Critical search & verification, groups compare a web search, a conversational model, and a citations-first engine, tracing every claim back to sources. Civic participation online tackles real local issues via democratic platforms (Decidim-style) and co-writes fair-use rules to avoid popularity contests. Create with care prototypes a mini-campaign and addresses authorship, derivative works, and the ethics of generated media. Safety by default is a rotating-station lab on cookies/consent, messaging-app settings, password managers, and metadata. A final strand, Prompting for truthfulness, builds guardrails against hallucination, bias, copyright exposure, and data leakage (roles/audiences/formats, “require citations,” retrieval for grounding). These flows were piloted in Italy and Spain and iterated to fit free vs. licensed tools and different learning contexts. Ethics and rights are kept inside the practice, not bolted on. Trainers collapse the EU ALTAI checklist into a few prompts—Who is affected? Where are the data from? Where is the human in the loop?—so dignity, transparency, oversight, robustness, and non-discrimination guide creation and participation in real time. Privacy is taught as a habit: unique passwords and managers, reading consent flows, understanding differences among messaging apps (E2E defaults, backups, metadata), and configuring settings on the spot. A simple “three promises” mnemonic helps: collect less, share consciously, retain briefly—illustrated with everyday cases like cookie choices and algorithmic recommendations.
The project scaffolds this method with a clear structure, roles, and outputs. Partners run six GymComp sessions × 3 hours (18 hours total) for 16 youth workers from Italy and Spain, using learning-by-doing, peer learning and “in-situation” practice. The same cohort meets for four online Workshops (2 hours each) to co-create a Learning Object (LO)—a multilingual (EN/IT/ES) bundle of materials, methodologies, and good practices for safe, ethical AI use in non-formal education. Arcipelago leads GymComp delivery; Fundación Esplai coordinates the Workshops and assembles the LO for open access via partner websites. Workshops are deliberately production-oriented: participants consolidate what worked in GymComp, document the steps, and turn procedures into reusable resources for peers. The emphasis is on transfer—from a facilitation routine tested in a youth club to a clearly documented method other youth workers can adopt the next day—rather than on showcasing specific apps. This keeps the project tool-agnostic: the process (plan → co-create → verify → attribute) matters more than brand names, which also helps maintain inclusion when free tiers shrink or features move behind paywalls. Across sessions, learning remains visible but light-touch: quick polls to set the pace, exit tickets aligned to DigComp descriptors, and short self-assessments to surface growth. For 3.1 tasks, rubrics look at intent clarity, licensing/attribution, audience fit, and transparency about AI assistance; for 4.2, evidence includes changed settings, stronger password hygiene, and a short “privacy story” (what changed and why). The result is evidence without bureaucracy—and behaviours that stick once devices are back in pockets. In the end, GymComp plus Workshops form a virtuous loop: situated practice → lightweight evidence → a shared, portable LO. This loop is what lets AIS4YWs keep its promises—quality youth work, inclusion and diversity, and a critical/ethical/safe use of AI—by turning European frameworks into everyday routines that are verifiable, transparent, and inclusive.
CHAPTER 3
Conclusion
Gym Comp shows that youth workers can turn European competence frameworks into repeatable, phone-friendly routines that fit real youth spaces. By mapping DigComp 2.2 (information and data literacy; participation; content creation; privacy and data protection) and LifeComp (communication, collaboration, empathy, self-regulation) onto short “circuits,” the project translates policy language into behaviours young people can practise and keep—verify first, create transparently, participate responsibly, protect data. This is exactly the spirit of the EU guidance our materials point to, and it keeps rights and well-being at the centre of practice. Two anchors make the approach durable beyond this chapter. First, safety-by-default is taught as habit, not theory: privacy principles (lawfulness, minimisation, storage limits) become concrete actions on the same devices youth carry—password managers, message-app settings, cookie choices, and plain-language explanations of GDPR, ePrivacy and DGA. This pairing of “why” with “how” is what changes behaviour. Second, trustworthy-by-design is simplified through ALTAI prompts—Who is affected? Where are the data from? Where is the human in the loop?—which youth workers can embed in any creative or participatory task.
The cross-country collaboration is a feature, not a backdrop. In Italy, Associazione Arcipelago APS and Lascò contributed facilitation know-how and ethics/prompting labs; coordination notes show how experts like Raffaele were woven into workshops and how local training lines are being planned for schools, libraries and rural areas. In Spain, Fundación Esplai has led the DigComp strands on participation and content creation, and coordinated next steps for shared training and open webinars. This Italy–Spain loop ensures the method is inclusive, tool-agnostic and ready to scale.For the wider field of non-formal education, Gym Comp offers a practical bridge: session objectives that foreground ethics alongside tools; prompting patterns that make model limits visible; and assessment moments that capture evidence without bureaucracy. These elements can slot into existing youth programmes, digital labs, and civic education modules, and they travel well across tools and bandwidth contexts.
CHAPTER 5
Conclusions
The examination of the relationships between the GDPR and the AI Act highlights concrete challenges that organizations will be required to face in the coming years. This entails the need to establish interdisciplinary teams with integrated expertise in privacy, AI, and regulatory compliance. Existing models and procedures will need to be updated to take into account not only data protection aspects, but also the specific features of AI, such as algorithm robustness or the possible presence of bias. Moreover, continuous monitoring systems will need to be implemented to verify the maintenance of compliance over time, since both sets of regulations require dynamic risk management. Only a pragmatic approach will be able to turn these regulatory challenges into opportunities for responsible innovation.
CHAPTER 1
Conclusion
The integration of AI into youth work is not simply a matter of adopting new tools—it challenges the very foundations of how we understand learning, agency, and the human relationship in education. While the AI4YW project and its predecessor demonstrated that AI can expand access, improve personalization, and foster engagement, they also revealed significant tensions and blind spots that require ongoing attention. Rebalancing power and agency One of the central questions emerging from the workshops is: Who controls the learning process when AI is involved? While generative tools empower users to create, simulate, and personalize, they also centralize technological power in the hands of opaque systems, often operated by private actors. Youth workers highlighted a need to preserve learner agency by ensuring AI is used as a scaffold—not a substitute—for critical thought, creativity, and collaboration. The GymComp approach, rooted in LIFEComp’s emphasis on personal and social competence, helped bring this into focus. By combining technical upskilling with transversal reflections, participants began to reframe AI not as a tool for efficiency alone, but as a space to question, negotiate, and co-create meaning. Moving beyond the "Toolbox" mentality A recurring insight was the importance of moving beyond a purely instrumental view of AI. The tendency to treat AI as a collection of apps risks fragmenting its ethical, social, and political implications. Instead, what emerged was a call to situate AI within the lived realities of youth—where issues of identity, inclusion, bias, and trust are constantly at stake.
Youth work, especially in non-formal education, is inherently relational. The best moments of the AI4YW process were not about what AI could do on behalf of educators, but about what it could enable in dialogue with them. These insights echo broader calls from the youth sector for a critical pedagogy of technology: one that puts human values, collective intelligence, and democratic oversight at its core.From Digital Competence to Digital MaturityDigComp 2.1 was essential in structuring the workshops, but the project experience showed that digital competence alone is not enough. What youth workers need is a form of digital maturity: the ability to make informed choices, critically evaluate systems, and support young people in navigating a complex technological landscape. This maturity cannot be acquired solely through checklists or tutorials—it requires time, dialogue, and reflective practice.In this sense, the GymComp model offered a useful prototype for how digital and transversal competences can be cultivated holistically. It recognized that understanding AI is not just about prompt engineering or data literacy—it’s about developing a mindset that combines ethical awareness, emotional intelligence, and adaptive learning. Toward a human-centered AI culture in Youth Work Finally, the project affirmed that integrating AI into youth work is not only about training or tools—it is about culture. The culture of youth work values participation, inclusion, care, and empowerment. For AI to be genuinely transformative, it must be subordinate to these values, not the other way around.
Chapter 5
Key content/discussion
The AI Act, on the other hand, defines a classification mechanism forartificial intelligence systems articulated into four categories of risk (unacceptable, high, limited, and minimal), each of which entails a different framework of obligations and responsibilities. This orientation emerges particularly clearly with the introduction, by the AI Act, of the Fundamental Rights Impact Assessment (FRIA), a preventive assessment tool that, while presenting points of contact with the DPIA under the GDPR, differs in scope and objectives. The DPIA, in fact, is specifically focused on risks related to the processing of personal data, whereas the FRIA adopts a broader perspective, taking into consideration the impact that artificial intelligence systems may have on the full range of fundamental rights guaranteed by the European legal order. While the DPIA focuses precisely on the privacy effects deriving from personal data processing, the FRIA broadens the perspective by considering a wider range of fundamental rights that may be affected by artificial intelligence systems, including the principle of non-discrimination, freedom of expression, human dignity, and the protection of minors. The two types of assessment make it possible to identify different risks that might escape an analysis focused exclusively on a single aspect: an AI system, for example, could prove compliant with data protection requirements while at the same time generating discriminatory effects, or it could respect fundamental rights without, however, ensuring adequate data security. To this end, the European legislator has introduced specific coordination mechanisms between the two tools: in cases where a high-risk AI system involves the processing of personal data, the FRIA can complement the elements already present in the DPIA, thus avoiding unnecessary overlaps.
Currently, the EU adopts a multi-level approach, in which the GDPR and the AI Act are combined with the DGA, the Data Act, the DSA, and the DMA, with the aim of ensuring substantial complementarity between the various regulatory instruments introduced over time. This strategy, based on the intention to foster technological innovation without neglecting the protection of European citizens’ fundamental rights, takes shape through an articulated set of interconnected rules that include, in addition to the regulations already mentioned, the Data Governance Act, the Data Act, the Digital Services Act, and the Digital Markets Act. In this complex regulatory framework, the relationship between the GDPR and the AI Act plays a leading role, since the use of artificial intelligence systems is closely linked to the management of large amounts of personal data, essential both for training algorithms and for their practical operation. The regulatory methodology envisaged both by the GDPR and by the AI Act is based on a risk-based approach which, while sharing common traits, takes different forms within their respective regulatory contexts.The GDPR, in fact, introduces a system of obligations that are graded according to the level of risk connected to processing, assigning the data controller the responsibility of assessing in advance the impact of operations on the rights and freedoms of data subjects through the instrument of the Data Protection Impact Assessment (DPIA).
My Digital Competence Self-Assessment (Based on GymComp)
2. Challenges: bias, inequality, and overdependence
Despite its promise, AI also presents substantial challenges that were discussed at length during workshops: Participants identified the risk of overreliance on AI outputs, especially by youth with limited critical thinking skills. Several youth workers shared concerns about young people treating AI responses as facts without questioning their source or limitations. Digital inequality emerged as a critical barrier: while some youth workers felt empowered by AI tools, others lacked the foundational digital skills to engage meaningfully with them. This divide risks widening existing social and educational inequalities unless proactive scaffolding is provided. The ethical risks of data misuse and lack of transparency in algorithmic decision-making were also flagged, especially when AI tools are used in sensitive contexts involving minors or marginalized groups. These concerns were mirrored in the AI4YouW pilot, where participants called for stronger guidance on privacy, clearer ethical use protocols, and more culturally adapted content for diverse learning communities
CHAPTER 3
Conclusion
Gym Comp shows that youth workers can turn European competence frameworks into repeatable, phone-friendly routines that fit real youth spaces. By mapping DigComp 2.2 (information and data literacy; participation; content creation; privacy and data protection) and LifeComp (communication, collaboration, empathy, self-regulation) onto short “circuits,” the project translates policy language into behaviours young people can practise and keep—verify first, create transparently, participate responsibly, protect data. This is exactly the spirit of the EU guidance our materials point to, and it keeps rights and well-being at the centre of practice. Two anchors make the approach durable beyond this chapter. First, safety-by-default is taught as habit, not theory: privacy principles (lawfulness, minimisation, storage limits) become concrete actions on the same devices youth carry—password managers, message-app settings, cookie choices, and plain-language explanations of GDPR, ePrivacy and DGA. This pairing of “why” with “how” is what changes behaviour. Second, trustworthy-by-design is simplified through ALTAI prompts—Who is affected? Where are the data from? Where is the human in the loop?—which youth workers can embed in any creative or participatory task.
The cross-country collaboration is a feature, not a backdrop. In Italy, Associazione Arcipelago APS and Lascò contributed facilitation know-how and ethics/prompting labs; coordination notes show how experts like Raffaele were woven into workshops and how local training lines are being planned for schools, libraries and rural areas. In Spain, Fundación Esplai has led the DigComp strands on participation and content creation, and coordinated next steps for shared training and open webinars. This Italy–Spain loop ensures the method is inclusive, tool-agnostic and ready to scale.For the wider field of non-formal education, Gym Comp offers a practical bridge: session objectives that foreground ethics alongside tools; prompting patterns that make model limits visible; and assessment moments that capture evidence without bureaucracy. These elements can slot into existing youth programmes, digital labs, and civic education modules, and they travel well across tools and bandwidth contexts.
AI Tools for Collaboration and Communication
These tools support group work, online facilitation, and interactive learning environments, with features enhanced by AI for organization or communication.
4) Tooling and accessibility
Training assets catalogue mainstream generators (Gemini, Copilot, ChatGPT, Canva) and accessibility add-ons (e.g., sign-language/voice tools) for inclusive delivery, reinforcing that the tool choice is secondary to the process: plan, co-create, verify, attribute. Session objectives are threefold: understanding AI-powered content types, selecting appropriate tools, and foregrounding ethical considerations. Live polls are used to gauge experience and expectations, then tailor depth and pacing.
CHAPTER 3
Reflexions
In recent years, the debate on Artificial Intelligence regulation has become central. AI is not a neutral technology: its decisions and uses can profoundly affect people’s lives, the market, and society. Why is regulation needed?
- Protection of fundamental rights – AI manages large amounts of personal data and can affect fundamental freedoms such as privacy, equality, and non-discrimination.
- Legal responsibility – Who is accountable if an AI system causes harm? It is necessary to define clear chains of responsibility, including developers, providers, and users.
- Promotion of trust and innovation – Clear regulation increases citizens’ trust and encourages safer investments by companies.
Ethical and social dimension of AI for critical and conscious use
- AI must be anthropocentric, serving human well-being and the common good.
- It must support democratic processes, fundamental rights, and the rule of law, avoiding risks such as mass surveillance or discriminatory bias.
Continuous and participatory reflection between developers, policymakers, and civil society is necessary to address ethical dilemmas and tensions between principles (e.g., security vs. individual freedom).
Introduction, definition and comptence construct
4. Competence development through GymComp
The GymComp methodology, structured around DigComp 2.1 and LIFEComp, played a key role in enabling self-assessment and capacity-building: Participants reflected on DigComp areas such as 1.1 Browsing and Searching, 2.3 Sharing and Collaborating, 3.1 Digital Content Creation, and 4.2 Protecting Personal Data, mapping their own confidence levels before and after the workshops. In line with LIFEComp, competences such as agency, emotional awareness, and learning-to-learn were embedded through role-play and scenario building. Several youth workers reported that this framework not only supported their own growth, but also gave them a language to talk about digital and transversal competences with the young people they support. As a result, over 75% of pilot participants stated they had developed new skills that would support AI adoption in their professional practice. Moreover, 84% indicated they would recommend the resources to peers, confirming both the pedagogical value and transferability of the approach.
2) Pedagogical stance: from frameworks to Youth Work
Our approach starts from learner realities and then maps back to the frameworks, an “inverted” design choice repeatedly raised in partner discussions. Trainers stress contextualising DigComp to communities, keeping the course practical, and recognising that certification matters only if it is perceived as meaningful by learners and employers. LifeComp anchors the relational side: self-regulation, flexibility, well-being; empathy, communication, collaboration; growth mindset, critical thinking and learning-to-learn. The slides structure each competence with awareness–understanding–action descriptors, which we translate into warm-ups (awareness), mini-inputs (understanding) and practice tasks (action). Italian partners contribute a concrete facilitation grammar—participant observation, identification of “bridge persons”, and explicit empathy techniques—so educators can surface the different “systems of representation” in a situation before acting. These notes guide facilitation choices in outreach contexts and are directly reused in Gym Comp circles.
Emerging competences in ethical use of AI and data
Youth Workers (YWs) play a key role in the conscious and safe adoption of artificial intelligence (AI) and data use in non-formal education. Within the overall DigCompEdu framework, there are specific competence indicators that can be useful in developing a digital and ethical culture in non-formal education for YWs. For further information on the system of skills useful for YWs in the use of AI in non-formal education, please refer to the table in the appendix.
AI Tools for Content Creation
AI-powered tools can assist educators and learners in creating content such as text, images, presentations, or videos. These tools can support creative expression, language learning, and digital storytelling.
Exploring AI in Daily Life
Ethical Dilemmas and AI
Tools for Assessment and Feedback
AI can support formative assessment by analyzing learners’ input, suggesting feedback, or providing personalised learning recommendations
CHAPTER 6
Conclusions
AI netiquette is not just a set of rules, but a way to build a harmonious relationship between humans and machines.
- Transparency builds trust.
- Respect for privacy ensures security.
- Fairness promotes equality.
- Responsibility and continuous improvement keep AI effective and ethical.
- The human touch ensures empathy and understanding.
As AI grows, these principles will guide us to use it responsibly and ethically. Welcoming AI with awareness means making it a partner, not just a tool. It is a way to build a future where AI benefits everyone.
Introduction, definition and comptence construct
AI Tools in Practice: A Critical Test
Educational Platforms Integrating AI
Some platforms combine learning management with AI features to personalize learning or provide adaptive feedback.
CHAPTER 5
Conclusions
The examination of the relationships between the GDPR and the AI Act highlights concrete challenges that organizations will be required to face in the coming years. This entails the need to establish interdisciplinary teams with integrated expertise in privacy, AI, and regulatory compliance. Existing models and procedures will need to be updated to take into account not only data protection aspects, but also the specific features of AI, such as algorithm robustness or the possible presence of bias. Moreover, continuous monitoring systems will need to be implemented to verify the maintenance of compliance over time, since both sets of regulations require dynamic risk management. Only a pragmatic approach will be able to turn these regulatory challenges into opportunities for responsible innovation.
9) Implementation notes from Italy and Spain
Coordination notes show how expertise improved delivery: inviting privacy and pedagogy experts; iterating on logistics; and planning cross-border workshops and open trainings. Teams stress that tool licensing and access constraints can undermine inclusion, so programmes must plan for equitable alternatives or institutional licences. Spanish partners document both the promise and the limits of digital participation (e.g., local voting processes turning into popularity contests), reinforcing the need for critical facilitation and bias awareness in civic tech activities..
LifeComp
DigiComp 2.3. - Participating in citizenship though digital technologies
CHAPTER 3
Conclusions
Building ethical AI requires multidisciplinary collaboration between:
- Developers and engineers, who understand the logic of models;
- Lawyers and legislators, who define the regulatory framework;
- Sociologists and ethicists, who assess social and cultural impact;
- End users, who provide real feedback on the use of technologies.
Ethics does not limit innovation: it makes it reliable and socially acceptable, creating trust and 'responsible competitiveness.' The ethical aspect must be incorporated into every phase of the AI system life cycle according to the principle of 'Ethics by Design.
7) Mitigating AI risks while creating value
Slides on “Modern AI” frame both excitement and risk: hallucinations, copyright exposure, data leaks, bias, and cost. The practice response is two-step: strengthen prompting discipline and introduce retrieval/grounding, always paired with human checking and explicit attribution. Partners showcased this in real tasks (e.g., drafting a workshop email safely and sourcing facts).
Tools for Ethical Reflection and Critical Thinking
In line with the project’s emphasis on responsibility and ethics, this sub-section highlights tools that support critical engagement with AI.
8) Assessment and evidence of learning
Formative assessment is continuous: quick polls on expectations, reflective exit tickets tied to DigiComp descriptors, and short self-assessments logged in the shared classroom space after sessions. This makes competence growth visible and actionable for both learners and facilitators. For 3.1 tasks, rubrics look at clarity of intent, ethical sourcing/licensing, suitability for audience, and transparency about AI assistance. For 4.2, learners must demonstrate changed settings, improved password hygiene, and an articulate “privacy story” (what they changed and why), not just recall of terms.
CHAPTER 5
Reflections
The concrete implementation of the provisions established by the GDPR and the AI Act raises significant technical and operational challenges, requiring the development of innovative solutions and the adoption of integrated approaches to regulatory compliance. Ensuring both compliance with data protection principles and adherence to the specific requirements established for artificial intelligence systems, in fact, entails designing technological architectures and organizational frameworks capable of jointly meeting the needs imposed by both sets of regulations.
Chapter 5
Key content/discussion
The AI Act, on the other hand, defines a classification mechanism forartificial intelligence systems articulated into four categories of risk (unacceptable, high, limited, and minimal), each of which entails a different framework of obligations and responsibilities. This orientation emerges particularly clearly with the introduction, by the AI Act, of the Fundamental Rights Impact Assessment (FRIA), a preventive assessment tool that, while presenting points of contact with the DPIA under the GDPR, differs in scope and objectives. The DPIA, in fact, is specifically focused on risks related to the processing of personal data, whereas the FRIA adopts a broader perspective, taking into consideration the impact that artificial intelligence systems may have on the full range of fundamental rights guaranteed by the European legal order. While the DPIA focuses precisely on the privacy effects deriving from personal data processing, the FRIA broadens the perspective by considering a wider range of fundamental rights that may be affected by artificial intelligence systems, including the principle of non-discrimination, freedom of expression, human dignity, and the protection of minors. The two types of assessment make it possible to identify different risks that might escape an analysis focused exclusively on a single aspect: an AI system, for example, could prove compliant with data protection requirements while at the same time generating discriminatory effects, or it could respect fundamental rights without, however, ensuring adequate data security. To this end, the European legislator has introduced specific coordination mechanisms between the two tools: in cases where a high-risk AI system involves the processing of personal data, the FRIA can complement the elements already present in the DPIA, thus avoiding unnecessary overlaps.
Currently, the EU adopts a multi-level approach, in which the GDPR and the AI Act are combined with the DGA, the Data Act, the DSA, and the DMA, with the aim of ensuring substantial complementarity between the various regulatory instruments introduced over time. This strategy, based on the intention to foster technological innovation without neglecting the protection of European citizens’ fundamental rights, takes shape through an articulated set of interconnected rules that include, in addition to the regulations already mentioned, the Data Governance Act, the Data Act, the Digital Services Act, and the Digital Markets Act. In this complex regulatory framework, the relationship between the GDPR and the AI Act plays a leading role, since the use of artificial intelligence systems is closely linked to the management of large amounts of personal data, essential both for training algorithms and for their practical operation. The regulatory methodology envisaged both by the GDPR and by the AI Act is based on a risk-based approach which, while sharing common traits, takes different forms within their respective regulatory contexts.The GDPR, in fact, introduces a system of obligations that are graded according to the level of risk connected to processing, assigning the data controller the responsibility of assessing in advance the impact of operations on the rights and freedoms of data subjects through the instrument of the Data Protection Impact Assessment (DPIA).
1) What Gym Comp targets
Gym Comp operationalises four DigComp 2.1 strands and connects them to LifeComp so sessions build both technical and relational capacity. Concretely, we focus on: (a) Information & data literacy (1.1–1.3) through AI-assisted search and source verification; (b) Digital participation (2.3) via civic platforms and youth-led advocacy; (c) Digital content creation (3.1) with prompt-driven writing, audio, image and video; and (d) Safety, privacy and data protection (4e.2) as “safety-by-default” routines in every activity. Session slides and meeting notes emphasise that 3.1 is not only “making things”, but doing so ethically and creatively with respect for authorship and rights
CHAPTER 1
Conclusion
AIS4YWs turns big frameworks into small, durable habits. With GymComp, youth workers practise verify → create, co-govern participation, and privacy by default until these moves become routine; with the Workshops, they package what works into a multilingual Learning Object other practitioners can adopt the next day. Together, these strands link DigComp 2.2 and LifeComp to everyday facilitation, keeping ethics, safety and inclusion inside the work rather than bolted on. This approach squarely serves the Erasmus+ priorities—quality and innovation in youth work and inclusion & diversity—by building a shared, cross-border method rather than a tool list. Roles are clear (Arcipelago coordinates GymComp and dissemination; Esplai leads monitoring, impact, and the LO) and the timeline embeds learning, evidence, and dissemination events that carry results beyond the cohort.
The project’s value also travels further: sessions and assets plug into existing youth programmes, civic education, libraries and community centres, while partner networks in Italy and Spain amplify reach and sustain practice after funding. In short, AIS4YWs offers a practical bridge from European policy to street-level education: repeatable circuits, lightweight evidence, and open resources that help young people create, verify, participate—and protect themselves and others—using AI responsibly If the sector holds to three commitments—verify then create, privacy by default, and co-governed participation—youth workers don’t need to be AI engineers to lead meaningful learning. They need a humane method, shared assets, and peers across Europe. AIS4YWs contributes all three.
6) Safety & privacy practices learners actually retain
Privacy sessions move beyond definitions into routine behaviours: unique passwords, managers, rotation; recognising metadata collection; reading consent flows; and understanding differences between messaging apps (end-to-end encryption defaults, backups, and metadata). Learners practise configuring settings and discuss trade-offs using concrete app examples. The legislative snapshot (GDPR, ePrivacy, Data Governance Act) is simplified into three learner promises: “collect less”, “share consciously”, “retain briefly”, explained through everyday cases such as cookies and algorithmic recommendations in social platforms.
4. Competence development through GymComp
The GymComp methodology, structured around DigComp 2.1 and LIFEComp, played a key role in enabling self-assessment and capacity-building: Participants reflected on DigComp areas such as 1.1 Browsing and Searching, 2.3 Sharing and Collaborating, 3.1 Digital Content Creation, and 4.2 Protecting Personal Data, mapping their own confidence levels before and after the workshops. In line with LIFEComp, competences such as agency, emotional awareness, and learning-to-learn were embedded through role-play and scenario building. Several youth workers reported that this framework not only supported their own growth, but also gave them a language to talk about digital and transversal competences with the young people they support. As a result, over 75% of pilot participants stated they had developed new skills that would support AI adoption in their professional practice. Moreover, 84% indicated they would recommend the resources to peers, confirming both the pedagogical value and transferability of the approach.
1. Opportunities: personalization, creativity, and inclusion
Workshop participants across Spain and Italy emphasized the potential of AI to personalize educational content, foster creativity, and open new avenues for inclusion: Generative tools such as ChatGPT enabled youth workers to design fictional personas and learning scenarios tailored to real-world contexts. For example, participants created personas such as Amina, a young legal assistant, for whom they developed a full learning path supported by AI-generated content. Brisk and MagicSchool were used to quickly generate quizzes, interactive exercises, and adapted materials. These tools were especially praised for helping educators simplify complex texts or create differentiated resources according to learners’ linguistic and cognitive needs. The impact matrices completed in the workshops highlighted that AI can facilitate access for young people with disabilities or language barriers—aligning with the LIFEComp dimensions of inclusiveness, learning-to-learn, and openness to diversity. This echoes the AI4YouW pilot results, where 84% of youth workers found the resources applicable to real-life youth work, and 78% reported improved ability to use AI in professional settings.
Objectives
- 1) Upskill Youth Workers on DigComp 2.2 areas 1.1, 1.2, 1.3, 2.3, 3.1, 4.2 and strengthen LifeComp (communication, collaboration, empathy, self-regulation) to foster inclusive online/offline environments.
- 2) Produce a digital Learning Obje ct (LO) with methods and tools for wise AI use in non-formal education, available in EN/IT/ES and freely downloadable from partner websites.
- 3) Bridge online and offline: trained Youth Workers act as “bridges,” helping young people transfer skills between digital and real-life settings, with special attention to those at risk of exclusion.
Activities
- GymComp: 6 online sessions × 3 hours (total 18 hours) to practise DigComp and LifeComp competences with self-assessment moments.
- Workshops: 4 online labs × 2 hours to assemble methods, tools and good practices
- GymComp curriculum (circuits, slides, checklists, facilitation notes, assessment rubrics) tested in Italy and Spain.
- Learning Object (EN/IT/ES) compiling methods and tools for safe, inclusive AI use in non-formal education; free download via partner sites.
Designing an AI-Enhanced Workshop
1. Opportunities: personalization, creativity, and inclusion
Workshop participants across Spain and Italy emphasized the potential of AI to personalize educational content, foster creativity, and open new avenues for inclusion: Generative tools such as ChatGPT enabled youth workers to design fictional personas and learning scenarios tailored to real-world contexts. For example, participants created personas such as Amina, a young legal assistant, for whom they developed a full learning path supported by AI-generated content. Brisk and MagicSchool were used to quickly generate quizzes, interactive exercises, and adapted materials. These tools were especially praised for helping educators simplify complex texts or create differentiated resources according to learners’ linguistic and cognitive needs. The impact matrices completed in the workshops highlighted that AI can facilitate access for young people with disabilities or language barriers—aligning with the LIFEComp dimensions of inclusiveness, learning-to-learn, and openness to diversity. This echoes the AI4YouW pilot results, where 84% of youth workers found the resources applicable to real-life youth work, and 78% reported improved ability to use AI in professional settings.
10) Why the LifeComp layer matters in AI practice
Beyond technique, Gym Comp relies on empathy and facilitation to reconnect young people, communities and institutions. Materials from the Italian outreach tradition formalise empathy as a methodological tool—suspending judgement, listening actively, and co-designing steps out of discomfort—so AI-enhanced activities remain human-centred and restorative. Bottom line: Gym Comp blends DigComp’s “can do” with LifeComp’s “how we do it together”. The materials translate EU guidance and partner experience into repeatable circuits where youth create, verify, participate—and protect themselves and others—using AI responsibly.
CHAPTER 3
Reflexions
Gym Comp works because it treats EU frameworks as means, not ends. The materials start from real youth-work frictions—privacy fatigue, tool access limits, “AI knows everything” myths—and translate DigComp/LifeComp into repeatable routines that learners can actually do under time pressure. That is the decisive shift: competences are not lists on a wall but behaviours embedded in short circuits—check the source, name the risk, set the control, reflect as a team. This framing mirrors LifeComp’s awareness–understanding–action arc and makes the “relational layer” visible rather than ornamental. A second takeaway is the centrality of safety-by-default. When privacy and security appear only as an add-on, they lose to convenience; when they are baked into every task (cookies, consent, passwords, messaging settings), learners start to narrate their own privacy stories and make concrete changes (e.g., enabling MFA, rotating passwords, choosing E2E defaults). The Italian privacy deck helps by turning abstract rules (lawful basis, purpose limitation, minimisation, storage limits) into short, memorizable prompts that participants can reuse across platforms. In practice, teaching the why of GDPR without the how of settings and habits is ineffective; Gym Comp keeps them coupled.. The sessions also surface an equity issue often ignored in AI trainings: access and licensing. Partners document how free tiers shrink and upload limits change, pushing learners toward paid plans. That matters in youth work, where the promise of “AI for everyone” collapses if activities depend on gated features. The honest response is not to lament but to design tool-agnostic exercises (verify, attribute, protect) and plan for institutional accounts or open alternatives when a licensed tool is pedagogically necessary. This keeps inclusion non-negotiable, not aspirational.
On the ethics and trust front, the ALTAI lens proves practical when collapsed into a few recurring questions: Who is affected? Where are the data from? Where is the human in the loop? What can go wrong, and how will we notice? In workshops, these prompts helped move from “AI is cool/scary” to auditable practice—traceable outputs, explicit attribution, and an agreed human review step. Rather than policing language, the checklists scaffold judgement and allow youth workers to justify choices to peers, funders, and participants.Still, AI risk literacy must be taught as technique and mindset. Slides on “modern AI hype” make a useful distinction: models can dazzle on unstructured data and still hallucinate confidently; learners need both prompting discipline and external grounding to resist that pull. In our view, the most robust habit taught here is “verify first, then create”: start from sources, constrain the model with them, and cite. This counters magical thinking without dampening creativity. The civic participation strand is where facilitation is most tested. Spanish discussions show how participatory platforms can drift into popularity contests and rivalry without guardrails. That is not a reason to abandon them; it is an argument for explicit norms, transparent data use, and bias-aware moderation designed with young people, not for them. Gym Comp’s choice to prototype participation rules inside the session is the right move: it treats youth as co-governors of the digital spaces they use.
Objectives
- 1) Upskill Youth Workers on DigComp 2.2 areas 1.1, 1.2, 1.3, 2.3, 3.1, 4.2 and strengthen LifeComp (communication, collaboration, empathy, self-regulation) to foster inclusive online/offline environments.
- 2) Produce a digital Learning Obje ct (LO) with methods and tools for wise AI use in non-formal education, available in EN/IT/ES and freely downloadable from partner websites.
- 3) Bridge online and offline: trained Youth Workers act as “bridges,” helping young people transfer skills between digital and real-life settings, with special attention to those at risk of exclusion.
Activities
- GymComp: 6 online sessions × 3 hours (total 18 hours) to practise DigComp and LifeComp competences with self-assessment moments.
- Workshops: 4 online labs × 2 hours to assemble methods, tools and good practices
- GymComp curriculum (circuits, slides, checklists, facilitation notes, assessment rubrics) tested in Italy and Spain.
- Learning Object (EN/IT/ES) compiling methods and tools for safe, inclusive AI use in non-formal education; free download via partner sites.
CHAPTER 3
Reflexions
In recent years, the debate on Artificial Intelligence regulation has become central. AI is not a neutral technology: its decisions and uses can profoundly affect people’s lives, the market, and society. Why is regulation needed?
- Protection of fundamental rights – AI manages large amounts of personal data and can affect fundamental freedoms such as privacy, equality, and non-discrimination.
- Legal responsibility – Who is accountable if an AI system causes harm? It is necessary to define clear chains of responsibility, including developers, providers, and users.
- Promotion of trust and innovation – Clear regulation increases citizens’ trust and encourages safer investments by companies.
Ethical and social dimension of AI for critical and conscious use
- AI must be anthropocentric, serving human well-being and the common good.
- It must support democratic processes, fundamental rights, and the rule of law, avoiding risks such as mass surveillance or discriminatory bias.
Continuous and participatory reflection between developers, policymakers, and civil society is necessary to address ethical dilemmas and tensions between principles (e.g., security vs. individual freedom).
CHAPTER 1
Conclusion
The integration of AI into youth work is not simply a matter of adopting new tools—it challenges the very foundations of how we understand learning, agency, and the human relationship in education. While the AI4YW project and its predecessor demonstrated that AI can expand access, improve personalization, and foster engagement, they also revealed significant tensions and blind spots that require ongoing attention. Rebalancing power and agency One of the central questions emerging from the workshops is: Who controls the learning process when AI is involved? While generative tools empower users to create, simulate, and personalize, they also centralize technological power in the hands of opaque systems, often operated by private actors. Youth workers highlighted a need to preserve learner agency by ensuring AI is used as a scaffold—not a substitute—for critical thought, creativity, and collaboration. The GymComp approach, rooted in LIFEComp’s emphasis on personal and social competence, helped bring this into focus. By combining technical upskilling with transversal reflections, participants began to reframe AI not as a tool for efficiency alone, but as a space to question, negotiate, and co-create meaning. Moving beyond the "Toolbox" mentality A recurring insight was the importance of moving beyond a purely instrumental view of AI. The tendency to treat AI as a collection of apps risks fragmenting its ethical, social, and political implications. Instead, what emerged was a call to situate AI within the lived realities of youth—where issues of identity, inclusion, bias, and trust are constantly at stake.
Youth work, especially in non-formal education, is inherently relational. The best moments of the AI4YW process were not about what AI could do on behalf of educators, but about what it could enable in dialogue with them. These insights echo broader calls from the youth sector for a critical pedagogy of technology: one that puts human values, collective intelligence, and democratic oversight at its core.From Digital Competence to Digital MaturityDigComp 2.1 was essential in structuring the workshops, but the project experience showed that digital competence alone is not enough. What youth workers need is a form of digital maturity: the ability to make informed choices, critically evaluate systems, and support young people in navigating a complex technological landscape. This maturity cannot be acquired solely through checklists or tutorials—it requires time, dialogue, and reflective practice.In this sense, the GymComp model offered a useful prototype for how digital and transversal competences can be cultivated holistically. It recognized that understanding AI is not just about prompt engineering or data literacy—it’s about developing a mindset that combines ethical awareness, emotional intelligence, and adaptive learning. Toward a human-centered AI culture in Youth Work Finally, the project affirmed that integrating AI into youth work is not only about training or tools—it is about culture. The culture of youth work values participation, inclusion, care, and empowerment. For AI to be genuinely transformative, it must be subordinate to these values, not the other way around.
3. Ethical reflection and responsible use
Ethics were not treated as an abstract add-on, but as a central pillar of all workshop activities. Drawing on the UNESCO framework and the LIFEComp focus on responsibility, collaboration, and critical thinking, the AI4YW project introduced a range of reflective tools: Participants engaged with examples such as the MIT “Moral Machine” experiment, discussing how AI might reinforce or challenge social biases. In collaborative settings, they identified situations in youth work where AI might undermine equity (e.g., automated profiling in job training programs), and co-created guidelines for responsible use. The importance of human oversight and the non-substitutive role of AI was stressed. AI was seen as a co-pilot—not a replacement—for the empathy, ethics, and relational work of youth professionals. This aligns with findings from AI4YW the pilot, where 78% of respondents affirmed that the OERs helped them recognize AI bias and promoted ethical awareness.
CHAPTER 3
Conclusions
Building ethical AI requires multidisciplinary collaboration between:
- Developers and engineers, who understand the logic of models;
- Lawyers and legislators, who define the regulatory framework;
- Sociologists and ethicists, who assess social and cultural impact;
- End users, who provide real feedback on the use of technologies.
Ethics does not limit innovation: it makes it reliable and socially acceptable, creating trust and 'responsible competitiveness.' The ethical aspect must be incorporated into every phase of the AI system life cycle according to the principle of 'Ethics by Design.
DigiCompt 3.1. - Digital content creation
LO-AI4YW-English version (original)
Fundación Esplai
Created on October 6, 2025
Start designing with a free template
Discover more than 1500 professional designs like these:
View
Teaching Challenge: Transform Your Classroom
View
Frayer Model
View
Math Calculations
View
Interactive QR Code Generator
View
Piñata Challenge
View
Interactive Scoreboard
View
Interactive Bingo
Explore all templates
Transcript
ABOUT THIS LEARNING OBJECT
AIS4YW is an Erasmus+ KA210-YOU small-scale partnership that strengthens the quality and innovation of youth work through hands-on, ethical use of generative AI in non-formal education. The 20-month project runs from 4 March 2024 to 3 November 2025. It is led by Associazione Arcipelago APS (Italy) with Fundación Esplai, Ciudadanía Comprometida (Spain) as partner. Priorities include improving youth-work quality and promoting inclusion and diversity. AIS4YW responds to the urgent need for youth workers (YWs) to understand how generative AI works, what opportunities it offers, which risks it entails, and how to deploy it critically, ethically, and safely with young and very young people.
The project is built as a community of practice between Italy and Spain to share methods and resources for professional, conscious AI use in non-formal learning. Direct participants are 16 youth workers (Italy & Spain) engaged through learning by doing, peer learning and “in-situation” practice. The content you will see below is the result of the exploration and practice carried out over the past months, which we hope will be useful to your Youth Work daily practice.
Table of contets
Chapter 1 - Introduction
Chapter 2 - Impact of AI in non-formal education
Chapter 3 - GymComp: Generative AI for Non-Formal Education
Chapter 4 - Ethics of AI Systems
Chapter 5 - Privacy & IA
Chapter 6 - Netiquette and Artificial Intelligence
This table of contents should allow you to easily navigate through the contents of this Learning Object. Browse through the chapters and find interactions with the content through images and icons.
Chapter 7 - Tools and practical examples
References
Chapter 1 - Introduction
AIS4YW is an Erasmus+ KA210-YOU small-scale partnership that strengthens the quality and innovation of youth work through hands-on, ethical use of generative AI in non-formal education. The 20-month project runs from 4 March 2024 to 3 November 2025. It is led by Associazione Arcipelago APS (Italy) with Fundación Esplai, Ciudadanía Comprometida (Spain) as partner. Priorities include improving youth-work quality and promoting inclusion and diversity. AIS4YW responds to the urgent need for youth workers (YWs) to understand how generative AI works, what opportunities it offers, which risks it entails, and how to deploy it critically, ethically, and safely with young and very young people. The project is built as a community of practice between Italy and Spain to share methods and resources for professional, conscious AI use in non-formal learning. Direct participants are 16 youth workers (Italy & Spain) engaged through learning by doing, peer learning and “in-situation” practice. Benefits extend to partner staff, local networks and broader European communities through the LO and dissemination events.
Objectives
Key content/discussion
Conclusion
Chapter 2 - Impact of AI in non-formal education
The impact of Artificial Intelligence (AI) on youth work and education is no longer a theoretical debate—it is a lived, dynamic reality. Building on the foundational work of the AI4YouW project, which piloted and validated open educational resources (OERs) with over 100 youth workers across Europe, the current AI4YW initiative expands that vision by actively engaging professionals in reflective, practical, and ethically grounded experimentation with generative AI tools. Through co-designed workshops held in Spain and Italy, educators explored how AI can reshape educational design, content creation, youth engagement, and professional development. These experiences were informed by global policy frameworks such as UNESCO’s 2023 guidance on generative AI in education, and conceptually anchored in the European Digital Competence Framework for Citizens (DigComp 2.1) and the LIFEComp framework for personal, social and learning-to-learn competences. Together, these frameworks provided the foundation for the GymComp methodology, a gamified and participatory process used to help youth workers reflect on and develop the competences needed for the responsible and effective use of AI. This chapter draws on diverse sources—workshop results, impact matrices, and pilot evaluations—to offer a grounded perspective on the opportunities, challenges, and transformative potential of AI. Rather than endorsing technological determinism, it advocates for a critically informed, inclusive, and human-centered approach to AI integration in youth work.
Key content/discussion
Conclusion
Chapter 2 - Practical examples
Example 3
Example 1
Example 2
Gamma – enables dynamic, visual presentation
Chatgpt Learning Experience Designer
Brisk Teaching – supports active learning through integrated micro-learning activities
Example 4
Example 5
Whimsical – facilitates the development of concept maps during group work
Mizou – supports the creation of interactive mock-ups for AI–persona interaction scenarios
Chapter 3 - GymComp: Generative AI for Non-Formal Education
This chapter introduces Gym Comp—a practical, hands-on approach to strengthening young people’s digital and relational capacities through short, repeatable training “circuits” that use generative AI in non-formal learning spaces. We translate key strands of the European competence frameworks—DigComp 2.2 (1.1–1.3 information and data literacy, 2.3 online participation, 3.1 digital content creation, 4.2 privacy and data protection) together with LifeComp’s communication, collaboration, empathy, and self-regulation—into activities that youth workers can run with minimal prep and maximum impact. Each circuit blends critical search and verification, prompt-driven creativity, and privacy-by-design routines aligned with GDPR, while openly addressing GenAI’s risks (bias, hallucination, copyright, safety) and showing how to mitigate them in everyday practice. The work presented here has been co-designed and piloted with partners in Italy and Spain. In Italy, the programme is curated by Associazione Arcipelago APS in collaboration with Lascò, Professor Raffaele Mele and Mattia Anicito, who together have adapted Gym Comp to community centres, youth clubs, libraries, and after-school contexts. In Spain, the roll-out and localisation are led by Fundación Esplai drawing on its long experience in digital inclusion and youth empowerment. This cross-country collaboration keeps the method grounded in real settings and diverse learner needs, while ensuring that the examples, scenarios, and protocols remain ethically sound and culturally relevant.
Key content/discussion
10
Reflexion
Conclusion
Chapter 3 - Appendices
Digicomp 2.3. - Participainting citizenship through digital technologies
DigiComp - Information and data literacy
Introdcution, definition and competence construct
LifeComp
DigiComp 4.2. - personal data proteccion and privacy
DigiComp 3.1. - Digital Content Creation
Chapter 4 - Ethics of AI Systems
The ethical dimension is one of the most relevant aspects of the impact of information technologies on social organization. The ethical aspect was little considered in the past, as people believed in the paradigm of 'self-regulation' of IT products, services, and activities, promoted for decades by the major tech giants. The scenario began to change thanks to two key factors: - Social awareness of the unforeseen negative effects of digital technology (e.g., information manipulation, loss of privacy, inequalities in access). - The global expansion of AI and, above all, of generative AI, which made evident the risks of a technology capable of making decisions or generating content with social, legal, and economic impacts. This changing scenario has made it necessary to raise ethical questions about the use of AI, even in fields outside of technology where AI has expanded and become essential in daily life, such as education (formal and non-formal).
Key content/discussion
Reflexion
Conclusion
Chapter 4 - Appendices
Ted Talks 1
Guiding Questions for Youth Workers on the Ethical Use of AI
Emerging competences in ethical use of AI and data for YWs
Ted Talks 2
Chapter 5 - Privacy and AI
The introduction and growing spread of artificial intelligence systems within today’s economic and social context have produced a significant change in the ways personal data are collected, processed, and used, raising regulatory issues of considerable complexity. The European legislator has chosen to respond to these challenges with a multi-level regulatory model, based on the interaction between the General Data Protection Regulation (GDPR) and the new Artificial Intelligence Regulation (AI Act), recently approved by the European Parliament as part of a broader EU digital strategy. Privacy protection exists only if it is guaranteed by laws, regulations, and directives that establish which data can be collected and in what ways.
Key content/discussion
Reflections
Conclusion
Chapter 6 - Netiquette and Artificial Intelligence
Netiquette is an English word that combines the English term network and the French word etiquette (good manners). It is a set of informal rules that regulate good user behavior on the web, especially in interacting with other users through resources such as newsgroups, mailing lists, forums, blogs, social networks, or email in general. Following the guidelines for acceptable behavior makes the internet a more pleasant place for all users. The way we interact with AI – and the way AI interacts with us – requires a new set of good manners
Key content/discussion
Conclusion
Chapter 7 - Tools and Practical Examples
This chapter presents a curated selection of tools, practice sheets, and tutorials that stem from the collaborative work of the AIS4YW project partnership. Developed in the context of non-formal education, the materials aim to support youth educators, trainers, and facilitators in integrating Artificial Intelligence (AI) into their educational practices in a thoughtful, inclusive, and ethically grounded way. The resources collected here reflect the core themes explored throughout the project, including the impact of AI on youth work, the development of digital and transversal competences (inspired by frameworks such as DigComp and LIFEComp), and the ethical and responsible use of digital technologies. The tools are aligned with these dimensions and were either co-created or selected by the project partners based on real needs identified during workshops, collaborative sessions, and local experimentation. Rather than offering an exhaustive inventory, this chapter highlights practical, ready-to-use resources — including AI-based tools, educational platforms, and examples of real-life applications — that can inspire educators in non-formal settings. The selected materials are adaptable to different contexts and learning environments and are designed to foster critical thinking, creativity, and learner empowerment. The chapter is structured in two sections: an overview of different toolkits and a collection of practice sheets on how to explore the use of AI with youth work. These sections aim to provide both conceptual clarity and hands-on guidance to those seeking to integrate AI in educational experiences beyond a non-formal context.
Chapter 7 - Toolkit overview
AI Tools for Collaboration and Communication
Educational Platforms Integrating AI
AI Tools for content creation
AI Tools for Ethical Reflection and Critical Thinking
AI Tools for Assessment and Feedback
Chapter 7 - Practice sheets
My Digital Competence Self-Assessment (Based on GymComp)
Ethical Dilemmas and AI
Exploring AI in Daily Life
Designing an AI-Enhanced Workshop
AI Tools in Practice: A Critical Test
References
Chapter 2
Chapter 3
References
Chapter 4
Chapter 5
References
Chapter 6
Get in touch with the project managers in each country: Italy -Arcipelago info@associazionearcipelago.com Spain - Fundación Esplai gporres@fundacionesplai.org
5) Ethics, rights and trustworthy-by-design
The materials adopt the EU’s ALTAI checklist to keep creation and participation “trustworthy”: respect fundamental rights (including minors), maintain human oversight, ensure technical robustness and safety, apply privacy-by-design/default, guarantee traceability and transparency, promote diversity/non-discrimination and consider social-environmental impacts. We convert these into checklists embedded in the circuits (e.g., “Who is affected?”, “Where are the data from?”, “Where is the human in the loop?”). Partners repeatedly note that learners must see how systems work—contrasting black-box with interpretable models, experimenting with simple training tools—to avoid “magical thinking” about AI.
DigiComp 4.2. - Personal data protection and privacy
Why AI ethics is crucial
AI is not just powerful software: it is a system capable of influencing decisions that once belonged exclusively to human beings. This implies direct consequences on fundamental rights such as: - Privacy, since AI processes enormous amounts of personal data; - Equal treatment, because uncontrolled algorithms can replicate or amplify social biases; - Safety, as errors or manipulations in AI systems can generate material or moral harm on a large scale. In recent years, it has become clear that 'relying on the common sense of technology creators' is not enough. What is needed are: - Clear regulatory frameworks, defining limits and responsibilities; - Shared ethical standards, guiding design, development, and implementation; - Transparency and audit mechanisms, enabling verification by independent entities.
CHAPTER 1
Conclusion
AIS4YWs turns big frameworks into small, durable habits. With GymComp, youth workers practise verify → create, co-govern participation, and privacy by default until these moves become routine; with the Workshops, they package what works into a multilingual Learning Object other practitioners can adopt the next day. Together, these strands link DigComp 2.2 and LifeComp to everyday facilitation, keeping ethics, safety and inclusion inside the work rather than bolted on. This approach squarely serves the Erasmus+ priorities—quality and innovation in youth work and inclusion & diversity—by building a shared, cross-border method rather than a tool list. Roles are clear (Arcipelago coordinates GymComp and dissemination; Esplai leads monitoring, impact, and the LO) and the timeline embeds learning, evidence, and dissemination events that carry results beyond the cohort.
The project’s value also travels further: sessions and assets plug into existing youth programmes, civic education, libraries and community centres, while partner networks in Italy and Spain amplify reach and sustain practice after funding. In short, AIS4YWs offers a practical bridge from European policy to street-level education: repeatable circuits, lightweight evidence, and open resources that help young people create, verify, participate—and protect themselves and others—using AI responsibly If the sector holds to three commitments—verify then create, privacy by default, and co-governed participation—youth workers don’t need to be AI engineers to lead meaningful learning. They need a humane method, shared assets, and peers across Europe. AIS4YWs contributes all three.
Transparency, fairness, and reliability
Reliability
- An AI system must function safely and consistently over time.
- Risk mitigation plans, regular updates, and independent audits must be provided.
In addition, further key aspects are necessary to ensure AI is reliable and safe:Three pillars are essential to ensure an AI system is ethical: Transparency
- Users and authorities must know how and why a system makes certain decisions.
- Opaque systems (black boxes) must be accompanied by explainable AI (XAI) tools.
- This aspect is fundamental for trust and the exercise of rights.
FairnessCHAPTER 5
Reflections
The concrete implementation of the provisions established by the GDPR and the AI Act raises significant technical and operational challenges, requiring the development of innovative solutions and the adoption of integrated approaches to regulatory compliance. Ensuring both compliance with data protection principles and adherence to the specific requirements established for artificial intelligence systems, in fact, entails designing technological architectures and organizational frameworks capable of jointly meeting the needs imposed by both sets of regulations.
Chapter 1
Key content/discussion
GymComp sits at the heart of AIS4YWs—a competence “gym” that turns the EU frameworks into short, repeatable, phone-friendly routines youth workers can run in real spaces with real young people. Rather than being a course “about AI models,” it’s a way to consistently do three things well: verify before you create, participate responsibly, and protect people and data. The design starts from concrete youth-work problems (myths like “AI knows everything,” privacy fatigue, licensing limits) and only then maps back to the frameworks, so competences don’t remain labels but become habits under constraint. In practice we work across DigComp 2.2 areas 1.1–1.3, 2.3, 3.1, 4.2 and weave in LifeComp’s social–emotional strands, so sessions build both technical and relational capacity. Each GymComp “circuit” is a brisk lab with clear timeboxes, rotating roles (scribe, checker, privacy lead), and a visible exit ticket tied to competence descriptors. In Critical search & verification, groups compare a web search, a conversational model, and a citations-first engine, tracing every claim back to sources. Civic participation online tackles real local issues via democratic platforms (Decidim-style) and co-writes fair-use rules to avoid popularity contests. Create with care prototypes a mini-campaign and addresses authorship, derivative works, and the ethics of generated media. Safety by default is a rotating-station lab on cookies/consent, messaging-app settings, password managers, and metadata. A final strand, Prompting for truthfulness, builds guardrails against hallucination, bias, copyright exposure, and data leakage (roles/audiences/formats, “require citations,” retrieval for grounding). These flows were piloted in Italy and Spain and iterated to fit free vs. licensed tools and different learning contexts. Ethics and rights are kept inside the practice, not bolted on. Trainers collapse the EU ALTAI checklist into a few prompts—Who is affected? Where are the data from? Where is the human in the loop?—so dignity, transparency, oversight, robustness, and non-discrimination guide creation and participation in real time. Privacy is taught as a habit: unique passwords and managers, reading consent flows, understanding differences among messaging apps (E2E defaults, backups, metadata), and configuring settings on the spot. A simple “three promises” mnemonic helps: collect less, share consciously, retain briefly—illustrated with everyday cases like cookie choices and algorithmic recommendations.
The project scaffolds this method with a clear structure, roles, and outputs. Partners run six GymComp sessions × 3 hours (18 hours total) for 16 youth workers from Italy and Spain, using learning-by-doing, peer learning and “in-situation” practice. The same cohort meets for four online Workshops (2 hours each) to co-create a Learning Object (LO)—a multilingual (EN/IT/ES) bundle of materials, methodologies, and good practices for safe, ethical AI use in non-formal education. Arcipelago leads GymComp delivery; Fundación Esplai coordinates the Workshops and assembles the LO for open access via partner websites. Workshops are deliberately production-oriented: participants consolidate what worked in GymComp, document the steps, and turn procedures into reusable resources for peers. The emphasis is on transfer—from a facilitation routine tested in a youth club to a clearly documented method other youth workers can adopt the next day—rather than on showcasing specific apps. This keeps the project tool-agnostic: the process (plan → co-create → verify → attribute) matters more than brand names, which also helps maintain inclusion when free tiers shrink or features move behind paywalls. Across sessions, learning remains visible but light-touch: quick polls to set the pace, exit tickets aligned to DigComp descriptors, and short self-assessments to surface growth. For 3.1 tasks, rubrics look at intent clarity, licensing/attribution, audience fit, and transparency about AI assistance; for 4.2, evidence includes changed settings, stronger password hygiene, and a short “privacy story” (what changed and why). The result is evidence without bureaucracy—and behaviours that stick once devices are back in pockets. In the end, GymComp plus Workshops form a virtuous loop: situated practice → lightweight evidence → a shared, portable LO. This loop is what lets AIS4YWs keep its promises—quality youth work, inclusion and diversity, and a critical/ethical/safe use of AI—by turning European frameworks into everyday routines that are verifiable, transparent, and inclusive.
CHAPTER 3
Reflexions
Gym Comp works because it treats EU frameworks as means, not ends. The materials start from real youth-work frictions—privacy fatigue, tool access limits, “AI knows everything” myths—and translate DigComp/LifeComp into repeatable routines that learners can actually do under time pressure. That is the decisive shift: competences are not lists on a wall but behaviours embedded in short circuits—check the source, name the risk, set the control, reflect as a team. This framing mirrors LifeComp’s awareness–understanding–action arc and makes the “relational layer” visible rather than ornamental. A second takeaway is the centrality of safety-by-default. When privacy and security appear only as an add-on, they lose to convenience; when they are baked into every task (cookies, consent, passwords, messaging settings), learners start to narrate their own privacy stories and make concrete changes (e.g., enabling MFA, rotating passwords, choosing E2E defaults). The Italian privacy deck helps by turning abstract rules (lawful basis, purpose limitation, minimisation, storage limits) into short, memorizable prompts that participants can reuse across platforms. In practice, teaching the why of GDPR without the how of settings and habits is ineffective; Gym Comp keeps them coupled.. The sessions also surface an equity issue often ignored in AI trainings: access and licensing. Partners document how free tiers shrink and upload limits change, pushing learners toward paid plans. That matters in youth work, where the promise of “AI for everyone” collapses if activities depend on gated features. The honest response is not to lament but to design tool-agnostic exercises (verify, attribute, protect) and plan for institutional accounts or open alternatives when a licensed tool is pedagogically necessary. This keeps inclusion non-negotiable, not aspirational.
On the ethics and trust front, the ALTAI lens proves practical when collapsed into a few recurring questions: Who is affected? Where are the data from? Where is the human in the loop? What can go wrong, and how will we notice? In workshops, these prompts helped move from “AI is cool/scary” to auditable practice—traceable outputs, explicit attribution, and an agreed human review step. Rather than policing language, the checklists scaffold judgement and allow youth workers to justify choices to peers, funders, and participants.Still, AI risk literacy must be taught as technique and mindset. Slides on “modern AI hype” make a useful distinction: models can dazzle on unstructured data and still hallucinate confidently; learners need both prompting discipline and external grounding to resist that pull. In our view, the most robust habit taught here is “verify first, then create”: start from sources, constrain the model with them, and cite. This counters magical thinking without dampening creativity. The civic participation strand is where facilitation is most tested. Spanish discussions show how participatory platforms can drift into popularity contests and rivalry without guardrails. That is not a reason to abandon them; it is an argument for explicit norms, transparent data use, and bias-aware moderation designed with young people, not for them. Gym Comp’s choice to prototype participation rules inside the session is the right move: it treats youth as co-governors of the digital spaces they use.
CHAPTER 6
Conclusions
AI netiquette is not just a set of rules, but a way to build a harmonious relationship between humans and machines.
- Transparency builds trust.
- Respect for privacy ensures security.
- Fairness promotes equality.
- Responsibility and continuous improvement keep AI effective and ethical.
- The human touch ensures empathy and understanding.
As AI grows, these principles will guide us to use it responsibly and ethically. Welcoming AI with awareness means making it a partner, not just a tool. It is a way to build a future where AI benefits everyone.2. Challenges: bias, inequality, and overdependence
Despite its promise, AI also presents substantial challenges that were discussed at length during workshops: Participants identified the risk of overreliance on AI outputs, especially by youth with limited critical thinking skills. Several youth workers shared concerns about young people treating AI responses as facts without questioning their source or limitations. Digital inequality emerged as a critical barrier: while some youth workers felt empowered by AI tools, others lacked the foundational digital skills to engage meaningfully with them. This divide risks widening existing social and educational inequalities unless proactive scaffolding is provided. The ethical risks of data misuse and lack of transparency in algorithmic decision-making were also flagged, especially when AI tools are used in sensitive contexts involving minors or marginalized groups. These concerns were mirrored in the AI4YouW pilot, where participants called for stronger guidance on privacy, clearer ethical use protocols, and more culturally adapted content for diverse learning communities
3) What learners do: core Gym Comp circuits
Circuit A — Critical search & verification (DigComp 1.1–1.3, 2.3; LifeComp: critical thinking). Learners compare answers from Google, a conversational model, and a citations-first engine, then trace claims back to sources. Partners found that tools which display sources transparently help demystify how results are produced and let learners judge credibility. Circuit B — Civic participation online (DigComp 2.3). Starting from local issues, groups explore democratic platforms (e.g., Decidim-style processes), reflect on pitfalls such as “popularity contests” in neighbourhood voting, and draft contribution guidelines that minimise bias and rivalry. Circuit C — Create with care (DigComp 3.1; LifeComp: collaboration & communication). Teams prototype a short campaign (poster, post, 30-sec video). Trainers introduce authorship, derivative works and the ethics of generated media; learners practise spotting AI-made vs human-made images and discuss fairness in contests where AI is used.
Circuit D — Safety by default (DigComp 4.2; LifeComp: self-regulation). A rotating station lab: manage cookies and consent banners; tune privacy on messaging apps; test password strength and managers; identify platform metadata and their implications. Slides summarise GDPR principles, ePrivacy scope, rights of the data subject, and pragmatic routines learners can apply immediately. Circuit E — Prompting for truthfulness. Learners see typical failure modes (hallucination, copyright, privacy leakage, bias, and cost trade-offs), then practise guardrails: specify role, audience and format; require citations; and use a retrieval step to ground outputs in provided materials. Each circuit includes timeboxes, roles (scribe, checker, privacy lead), and a visible exit ticket aligned with the competence descriptors. Italian and Spanish cohorts tested these flows in youth clubs and workshops, iterating based on learner feedback and the practical limits of free vs licensed tools.
Chapter 1
Key content/discussion
GymComp sits at the heart of AIS4YWs—a competence “gym” that turns the EU frameworks into short, repeatable, phone-friendly routines youth workers can run in real spaces with real young people. Rather than being a course “about AI models,” it’s a way to consistently do three things well: verify before you create, participate responsibly, and protect people and data. The design starts from concrete youth-work problems (myths like “AI knows everything,” privacy fatigue, licensing limits) and only then maps back to the frameworks, so competences don’t remain labels but become habits under constraint. In practice we work across DigComp 2.2 areas 1.1–1.3, 2.3, 3.1, 4.2 and weave in LifeComp’s social–emotional strands, so sessions build both technical and relational capacity. Each GymComp “circuit” is a brisk lab with clear timeboxes, rotating roles (scribe, checker, privacy lead), and a visible exit ticket tied to competence descriptors. In Critical search & verification, groups compare a web search, a conversational model, and a citations-first engine, tracing every claim back to sources. Civic participation online tackles real local issues via democratic platforms (Decidim-style) and co-writes fair-use rules to avoid popularity contests. Create with care prototypes a mini-campaign and addresses authorship, derivative works, and the ethics of generated media. Safety by default is a rotating-station lab on cookies/consent, messaging-app settings, password managers, and metadata. A final strand, Prompting for truthfulness, builds guardrails against hallucination, bias, copyright exposure, and data leakage (roles/audiences/formats, “require citations,” retrieval for grounding). These flows were piloted in Italy and Spain and iterated to fit free vs. licensed tools and different learning contexts. Ethics and rights are kept inside the practice, not bolted on. Trainers collapse the EU ALTAI checklist into a few prompts—Who is affected? Where are the data from? Where is the human in the loop?—so dignity, transparency, oversight, robustness, and non-discrimination guide creation and participation in real time. Privacy is taught as a habit: unique passwords and managers, reading consent flows, understanding differences among messaging apps (E2E defaults, backups, metadata), and configuring settings on the spot. A simple “three promises” mnemonic helps: collect less, share consciously, retain briefly—illustrated with everyday cases like cookie choices and algorithmic recommendations.
The project scaffolds this method with a clear structure, roles, and outputs. Partners run six GymComp sessions × 3 hours (18 hours total) for 16 youth workers from Italy and Spain, using learning-by-doing, peer learning and “in-situation” practice. The same cohort meets for four online Workshops (2 hours each) to co-create a Learning Object (LO)—a multilingual (EN/IT/ES) bundle of materials, methodologies, and good practices for safe, ethical AI use in non-formal education. Arcipelago leads GymComp delivery; Fundación Esplai coordinates the Workshops and assembles the LO for open access via partner websites. Workshops are deliberately production-oriented: participants consolidate what worked in GymComp, document the steps, and turn procedures into reusable resources for peers. The emphasis is on transfer—from a facilitation routine tested in a youth club to a clearly documented method other youth workers can adopt the next day—rather than on showcasing specific apps. This keeps the project tool-agnostic: the process (plan → co-create → verify → attribute) matters more than brand names, which also helps maintain inclusion when free tiers shrink or features move behind paywalls. Across sessions, learning remains visible but light-touch: quick polls to set the pace, exit tickets aligned to DigComp descriptors, and short self-assessments to surface growth. For 3.1 tasks, rubrics look at intent clarity, licensing/attribution, audience fit, and transparency about AI assistance; for 4.2, evidence includes changed settings, stronger password hygiene, and a short “privacy story” (what changed and why). The result is evidence without bureaucracy—and behaviours that stick once devices are back in pockets. In the end, GymComp plus Workshops form a virtuous loop: situated practice → lightweight evidence → a shared, portable LO. This loop is what lets AIS4YWs keep its promises—quality youth work, inclusion and diversity, and a critical/ethical/safe use of AI—by turning European frameworks into everyday routines that are verifiable, transparent, and inclusive.
CHAPTER 3
Conclusion
Gym Comp shows that youth workers can turn European competence frameworks into repeatable, phone-friendly routines that fit real youth spaces. By mapping DigComp 2.2 (information and data literacy; participation; content creation; privacy and data protection) and LifeComp (communication, collaboration, empathy, self-regulation) onto short “circuits,” the project translates policy language into behaviours young people can practise and keep—verify first, create transparently, participate responsibly, protect data. This is exactly the spirit of the EU guidance our materials point to, and it keeps rights and well-being at the centre of practice. Two anchors make the approach durable beyond this chapter. First, safety-by-default is taught as habit, not theory: privacy principles (lawfulness, minimisation, storage limits) become concrete actions on the same devices youth carry—password managers, message-app settings, cookie choices, and plain-language explanations of GDPR, ePrivacy and DGA. This pairing of “why” with “how” is what changes behaviour. Second, trustworthy-by-design is simplified through ALTAI prompts—Who is affected? Where are the data from? Where is the human in the loop?—which youth workers can embed in any creative or participatory task.
The cross-country collaboration is a feature, not a backdrop. In Italy, Associazione Arcipelago APS and Lascò contributed facilitation know-how and ethics/prompting labs; coordination notes show how experts like Raffaele were woven into workshops and how local training lines are being planned for schools, libraries and rural areas. In Spain, Fundación Esplai has led the DigComp strands on participation and content creation, and coordinated next steps for shared training and open webinars. This Italy–Spain loop ensures the method is inclusive, tool-agnostic and ready to scale.For the wider field of non-formal education, Gym Comp offers a practical bridge: session objectives that foreground ethics alongside tools; prompting patterns that make model limits visible; and assessment moments that capture evidence without bureaucracy. These elements can slot into existing youth programmes, digital labs, and civic education modules, and they travel well across tools and bandwidth contexts.
CHAPTER 5
Conclusions
The examination of the relationships between the GDPR and the AI Act highlights concrete challenges that organizations will be required to face in the coming years. This entails the need to establish interdisciplinary teams with integrated expertise in privacy, AI, and regulatory compliance. Existing models and procedures will need to be updated to take into account not only data protection aspects, but also the specific features of AI, such as algorithm robustness or the possible presence of bias. Moreover, continuous monitoring systems will need to be implemented to verify the maintenance of compliance over time, since both sets of regulations require dynamic risk management. Only a pragmatic approach will be able to turn these regulatory challenges into opportunities for responsible innovation.
CHAPTER 1
Conclusion
The integration of AI into youth work is not simply a matter of adopting new tools—it challenges the very foundations of how we understand learning, agency, and the human relationship in education. While the AI4YW project and its predecessor demonstrated that AI can expand access, improve personalization, and foster engagement, they also revealed significant tensions and blind spots that require ongoing attention. Rebalancing power and agency One of the central questions emerging from the workshops is: Who controls the learning process when AI is involved? While generative tools empower users to create, simulate, and personalize, they also centralize technological power in the hands of opaque systems, often operated by private actors. Youth workers highlighted a need to preserve learner agency by ensuring AI is used as a scaffold—not a substitute—for critical thought, creativity, and collaboration. The GymComp approach, rooted in LIFEComp’s emphasis on personal and social competence, helped bring this into focus. By combining technical upskilling with transversal reflections, participants began to reframe AI not as a tool for efficiency alone, but as a space to question, negotiate, and co-create meaning. Moving beyond the "Toolbox" mentality A recurring insight was the importance of moving beyond a purely instrumental view of AI. The tendency to treat AI as a collection of apps risks fragmenting its ethical, social, and political implications. Instead, what emerged was a call to situate AI within the lived realities of youth—where issues of identity, inclusion, bias, and trust are constantly at stake.
Youth work, especially in non-formal education, is inherently relational. The best moments of the AI4YW process were not about what AI could do on behalf of educators, but about what it could enable in dialogue with them. These insights echo broader calls from the youth sector for a critical pedagogy of technology: one that puts human values, collective intelligence, and democratic oversight at its core.From Digital Competence to Digital MaturityDigComp 2.1 was essential in structuring the workshops, but the project experience showed that digital competence alone is not enough. What youth workers need is a form of digital maturity: the ability to make informed choices, critically evaluate systems, and support young people in navigating a complex technological landscape. This maturity cannot be acquired solely through checklists or tutorials—it requires time, dialogue, and reflective practice.In this sense, the GymComp model offered a useful prototype for how digital and transversal competences can be cultivated holistically. It recognized that understanding AI is not just about prompt engineering or data literacy—it’s about developing a mindset that combines ethical awareness, emotional intelligence, and adaptive learning. Toward a human-centered AI culture in Youth Work Finally, the project affirmed that integrating AI into youth work is not only about training or tools—it is about culture. The culture of youth work values participation, inclusion, care, and empowerment. For AI to be genuinely transformative, it must be subordinate to these values, not the other way around.
Chapter 5
Key content/discussion
The AI Act, on the other hand, defines a classification mechanism forartificial intelligence systems articulated into four categories of risk (unacceptable, high, limited, and minimal), each of which entails a different framework of obligations and responsibilities. This orientation emerges particularly clearly with the introduction, by the AI Act, of the Fundamental Rights Impact Assessment (FRIA), a preventive assessment tool that, while presenting points of contact with the DPIA under the GDPR, differs in scope and objectives. The DPIA, in fact, is specifically focused on risks related to the processing of personal data, whereas the FRIA adopts a broader perspective, taking into consideration the impact that artificial intelligence systems may have on the full range of fundamental rights guaranteed by the European legal order. While the DPIA focuses precisely on the privacy effects deriving from personal data processing, the FRIA broadens the perspective by considering a wider range of fundamental rights that may be affected by artificial intelligence systems, including the principle of non-discrimination, freedom of expression, human dignity, and the protection of minors. The two types of assessment make it possible to identify different risks that might escape an analysis focused exclusively on a single aspect: an AI system, for example, could prove compliant with data protection requirements while at the same time generating discriminatory effects, or it could respect fundamental rights without, however, ensuring adequate data security. To this end, the European legislator has introduced specific coordination mechanisms between the two tools: in cases where a high-risk AI system involves the processing of personal data, the FRIA can complement the elements already present in the DPIA, thus avoiding unnecessary overlaps.
Currently, the EU adopts a multi-level approach, in which the GDPR and the AI Act are combined with the DGA, the Data Act, the DSA, and the DMA, with the aim of ensuring substantial complementarity between the various regulatory instruments introduced over time. This strategy, based on the intention to foster technological innovation without neglecting the protection of European citizens’ fundamental rights, takes shape through an articulated set of interconnected rules that include, in addition to the regulations already mentioned, the Data Governance Act, the Data Act, the Digital Services Act, and the Digital Markets Act. In this complex regulatory framework, the relationship between the GDPR and the AI Act plays a leading role, since the use of artificial intelligence systems is closely linked to the management of large amounts of personal data, essential both for training algorithms and for their practical operation. The regulatory methodology envisaged both by the GDPR and by the AI Act is based on a risk-based approach which, while sharing common traits, takes different forms within their respective regulatory contexts.The GDPR, in fact, introduces a system of obligations that are graded according to the level of risk connected to processing, assigning the data controller the responsibility of assessing in advance the impact of operations on the rights and freedoms of data subjects through the instrument of the Data Protection Impact Assessment (DPIA).
My Digital Competence Self-Assessment (Based on GymComp)
2. Challenges: bias, inequality, and overdependence
Despite its promise, AI also presents substantial challenges that were discussed at length during workshops: Participants identified the risk of overreliance on AI outputs, especially by youth with limited critical thinking skills. Several youth workers shared concerns about young people treating AI responses as facts without questioning their source or limitations. Digital inequality emerged as a critical barrier: while some youth workers felt empowered by AI tools, others lacked the foundational digital skills to engage meaningfully with them. This divide risks widening existing social and educational inequalities unless proactive scaffolding is provided. The ethical risks of data misuse and lack of transparency in algorithmic decision-making were also flagged, especially when AI tools are used in sensitive contexts involving minors or marginalized groups. These concerns were mirrored in the AI4YouW pilot, where participants called for stronger guidance on privacy, clearer ethical use protocols, and more culturally adapted content for diverse learning communities
CHAPTER 3
Conclusion
Gym Comp shows that youth workers can turn European competence frameworks into repeatable, phone-friendly routines that fit real youth spaces. By mapping DigComp 2.2 (information and data literacy; participation; content creation; privacy and data protection) and LifeComp (communication, collaboration, empathy, self-regulation) onto short “circuits,” the project translates policy language into behaviours young people can practise and keep—verify first, create transparently, participate responsibly, protect data. This is exactly the spirit of the EU guidance our materials point to, and it keeps rights and well-being at the centre of practice. Two anchors make the approach durable beyond this chapter. First, safety-by-default is taught as habit, not theory: privacy principles (lawfulness, minimisation, storage limits) become concrete actions on the same devices youth carry—password managers, message-app settings, cookie choices, and plain-language explanations of GDPR, ePrivacy and DGA. This pairing of “why” with “how” is what changes behaviour. Second, trustworthy-by-design is simplified through ALTAI prompts—Who is affected? Where are the data from? Where is the human in the loop?—which youth workers can embed in any creative or participatory task.
The cross-country collaboration is a feature, not a backdrop. In Italy, Associazione Arcipelago APS and Lascò contributed facilitation know-how and ethics/prompting labs; coordination notes show how experts like Raffaele were woven into workshops and how local training lines are being planned for schools, libraries and rural areas. In Spain, Fundación Esplai has led the DigComp strands on participation and content creation, and coordinated next steps for shared training and open webinars. This Italy–Spain loop ensures the method is inclusive, tool-agnostic and ready to scale.For the wider field of non-formal education, Gym Comp offers a practical bridge: session objectives that foreground ethics alongside tools; prompting patterns that make model limits visible; and assessment moments that capture evidence without bureaucracy. These elements can slot into existing youth programmes, digital labs, and civic education modules, and they travel well across tools and bandwidth contexts.
AI Tools for Collaboration and Communication
These tools support group work, online facilitation, and interactive learning environments, with features enhanced by AI for organization or communication.
4) Tooling and accessibility
Training assets catalogue mainstream generators (Gemini, Copilot, ChatGPT, Canva) and accessibility add-ons (e.g., sign-language/voice tools) for inclusive delivery, reinforcing that the tool choice is secondary to the process: plan, co-create, verify, attribute. Session objectives are threefold: understanding AI-powered content types, selecting appropriate tools, and foregrounding ethical considerations. Live polls are used to gauge experience and expectations, then tailor depth and pacing.
CHAPTER 3
Reflexions
In recent years, the debate on Artificial Intelligence regulation has become central. AI is not a neutral technology: its decisions and uses can profoundly affect people’s lives, the market, and society. Why is regulation needed?
- Protection of fundamental rights – AI manages large amounts of personal data and can affect fundamental freedoms such as privacy, equality, and non-discrimination.
- Legal responsibility – Who is accountable if an AI system causes harm? It is necessary to define clear chains of responsibility, including developers, providers, and users.
- Promotion of trust and innovation – Clear regulation increases citizens’ trust and encourages safer investments by companies.
Ethical and social dimension of AI for critical and conscious use- AI must be anthropocentric, serving human well-being and the common good.
- It must support democratic processes, fundamental rights, and the rule of law, avoiding risks such as mass surveillance or discriminatory bias.
Continuous and participatory reflection between developers, policymakers, and civil society is necessary to address ethical dilemmas and tensions between principles (e.g., security vs. individual freedom).Introduction, definition and comptence construct
4. Competence development through GymComp
The GymComp methodology, structured around DigComp 2.1 and LIFEComp, played a key role in enabling self-assessment and capacity-building: Participants reflected on DigComp areas such as 1.1 Browsing and Searching, 2.3 Sharing and Collaborating, 3.1 Digital Content Creation, and 4.2 Protecting Personal Data, mapping their own confidence levels before and after the workshops. In line with LIFEComp, competences such as agency, emotional awareness, and learning-to-learn were embedded through role-play and scenario building. Several youth workers reported that this framework not only supported their own growth, but also gave them a language to talk about digital and transversal competences with the young people they support. As a result, over 75% of pilot participants stated they had developed new skills that would support AI adoption in their professional practice. Moreover, 84% indicated they would recommend the resources to peers, confirming both the pedagogical value and transferability of the approach.
2) Pedagogical stance: from frameworks to Youth Work
Our approach starts from learner realities and then maps back to the frameworks, an “inverted” design choice repeatedly raised in partner discussions. Trainers stress contextualising DigComp to communities, keeping the course practical, and recognising that certification matters only if it is perceived as meaningful by learners and employers. LifeComp anchors the relational side: self-regulation, flexibility, well-being; empathy, communication, collaboration; growth mindset, critical thinking and learning-to-learn. The slides structure each competence with awareness–understanding–action descriptors, which we translate into warm-ups (awareness), mini-inputs (understanding) and practice tasks (action). Italian partners contribute a concrete facilitation grammar—participant observation, identification of “bridge persons”, and explicit empathy techniques—so educators can surface the different “systems of representation” in a situation before acting. These notes guide facilitation choices in outreach contexts and are directly reused in Gym Comp circles.
Emerging competences in ethical use of AI and data
Youth Workers (YWs) play a key role in the conscious and safe adoption of artificial intelligence (AI) and data use in non-formal education. Within the overall DigCompEdu framework, there are specific competence indicators that can be useful in developing a digital and ethical culture in non-formal education for YWs. For further information on the system of skills useful for YWs in the use of AI in non-formal education, please refer to the table in the appendix.
AI Tools for Content Creation
AI-powered tools can assist educators and learners in creating content such as text, images, presentations, or videos. These tools can support creative expression, language learning, and digital storytelling.
Exploring AI in Daily Life
Ethical Dilemmas and AI
Tools for Assessment and Feedback
AI can support formative assessment by analyzing learners’ input, suggesting feedback, or providing personalised learning recommendations
CHAPTER 6
Conclusions
AI netiquette is not just a set of rules, but a way to build a harmonious relationship between humans and machines.
- Transparency builds trust.
- Respect for privacy ensures security.
- Fairness promotes equality.
- Responsibility and continuous improvement keep AI effective and ethical.
- The human touch ensures empathy and understanding.
As AI grows, these principles will guide us to use it responsibly and ethically. Welcoming AI with awareness means making it a partner, not just a tool. It is a way to build a future where AI benefits everyone.Introduction, definition and comptence construct
AI Tools in Practice: A Critical Test
Educational Platforms Integrating AI
Some platforms combine learning management with AI features to personalize learning or provide adaptive feedback.
CHAPTER 5
Conclusions
The examination of the relationships between the GDPR and the AI Act highlights concrete challenges that organizations will be required to face in the coming years. This entails the need to establish interdisciplinary teams with integrated expertise in privacy, AI, and regulatory compliance. Existing models and procedures will need to be updated to take into account not only data protection aspects, but also the specific features of AI, such as algorithm robustness or the possible presence of bias. Moreover, continuous monitoring systems will need to be implemented to verify the maintenance of compliance over time, since both sets of regulations require dynamic risk management. Only a pragmatic approach will be able to turn these regulatory challenges into opportunities for responsible innovation.
9) Implementation notes from Italy and Spain
Coordination notes show how expertise improved delivery: inviting privacy and pedagogy experts; iterating on logistics; and planning cross-border workshops and open trainings. Teams stress that tool licensing and access constraints can undermine inclusion, so programmes must plan for equitable alternatives or institutional licences. Spanish partners document both the promise and the limits of digital participation (e.g., local voting processes turning into popularity contests), reinforcing the need for critical facilitation and bias awareness in civic tech activities..
LifeComp
DigiComp 2.3. - Participating in citizenship though digital technologies
CHAPTER 3
Conclusions
Building ethical AI requires multidisciplinary collaboration between:
- Developers and engineers, who understand the logic of models;
- Lawyers and legislators, who define the regulatory framework;
- Sociologists and ethicists, who assess social and cultural impact;
- End users, who provide real feedback on the use of technologies.
Ethics does not limit innovation: it makes it reliable and socially acceptable, creating trust and 'responsible competitiveness.' The ethical aspect must be incorporated into every phase of the AI system life cycle according to the principle of 'Ethics by Design.7) Mitigating AI risks while creating value
Slides on “Modern AI” frame both excitement and risk: hallucinations, copyright exposure, data leaks, bias, and cost. The practice response is two-step: strengthen prompting discipline and introduce retrieval/grounding, always paired with human checking and explicit attribution. Partners showcased this in real tasks (e.g., drafting a workshop email safely and sourcing facts).
Tools for Ethical Reflection and Critical Thinking
In line with the project’s emphasis on responsibility and ethics, this sub-section highlights tools that support critical engagement with AI.
8) Assessment and evidence of learning
Formative assessment is continuous: quick polls on expectations, reflective exit tickets tied to DigiComp descriptors, and short self-assessments logged in the shared classroom space after sessions. This makes competence growth visible and actionable for both learners and facilitators. For 3.1 tasks, rubrics look at clarity of intent, ethical sourcing/licensing, suitability for audience, and transparency about AI assistance. For 4.2, learners must demonstrate changed settings, improved password hygiene, and an articulate “privacy story” (what they changed and why), not just recall of terms.
CHAPTER 5
Reflections
The concrete implementation of the provisions established by the GDPR and the AI Act raises significant technical and operational challenges, requiring the development of innovative solutions and the adoption of integrated approaches to regulatory compliance. Ensuring both compliance with data protection principles and adherence to the specific requirements established for artificial intelligence systems, in fact, entails designing technological architectures and organizational frameworks capable of jointly meeting the needs imposed by both sets of regulations.
Chapter 5
Key content/discussion
The AI Act, on the other hand, defines a classification mechanism forartificial intelligence systems articulated into four categories of risk (unacceptable, high, limited, and minimal), each of which entails a different framework of obligations and responsibilities. This orientation emerges particularly clearly with the introduction, by the AI Act, of the Fundamental Rights Impact Assessment (FRIA), a preventive assessment tool that, while presenting points of contact with the DPIA under the GDPR, differs in scope and objectives. The DPIA, in fact, is specifically focused on risks related to the processing of personal data, whereas the FRIA adopts a broader perspective, taking into consideration the impact that artificial intelligence systems may have on the full range of fundamental rights guaranteed by the European legal order. While the DPIA focuses precisely on the privacy effects deriving from personal data processing, the FRIA broadens the perspective by considering a wider range of fundamental rights that may be affected by artificial intelligence systems, including the principle of non-discrimination, freedom of expression, human dignity, and the protection of minors. The two types of assessment make it possible to identify different risks that might escape an analysis focused exclusively on a single aspect: an AI system, for example, could prove compliant with data protection requirements while at the same time generating discriminatory effects, or it could respect fundamental rights without, however, ensuring adequate data security. To this end, the European legislator has introduced specific coordination mechanisms between the two tools: in cases where a high-risk AI system involves the processing of personal data, the FRIA can complement the elements already present in the DPIA, thus avoiding unnecessary overlaps.
Currently, the EU adopts a multi-level approach, in which the GDPR and the AI Act are combined with the DGA, the Data Act, the DSA, and the DMA, with the aim of ensuring substantial complementarity between the various regulatory instruments introduced over time. This strategy, based on the intention to foster technological innovation without neglecting the protection of European citizens’ fundamental rights, takes shape through an articulated set of interconnected rules that include, in addition to the regulations already mentioned, the Data Governance Act, the Data Act, the Digital Services Act, and the Digital Markets Act. In this complex regulatory framework, the relationship between the GDPR and the AI Act plays a leading role, since the use of artificial intelligence systems is closely linked to the management of large amounts of personal data, essential both for training algorithms and for their practical operation. The regulatory methodology envisaged both by the GDPR and by the AI Act is based on a risk-based approach which, while sharing common traits, takes different forms within their respective regulatory contexts.The GDPR, in fact, introduces a system of obligations that are graded according to the level of risk connected to processing, assigning the data controller the responsibility of assessing in advance the impact of operations on the rights and freedoms of data subjects through the instrument of the Data Protection Impact Assessment (DPIA).
1) What Gym Comp targets
Gym Comp operationalises four DigComp 2.1 strands and connects them to LifeComp so sessions build both technical and relational capacity. Concretely, we focus on: (a) Information & data literacy (1.1–1.3) through AI-assisted search and source verification; (b) Digital participation (2.3) via civic platforms and youth-led advocacy; (c) Digital content creation (3.1) with prompt-driven writing, audio, image and video; and (d) Safety, privacy and data protection (4e.2) as “safety-by-default” routines in every activity. Session slides and meeting notes emphasise that 3.1 is not only “making things”, but doing so ethically and creatively with respect for authorship and rights
CHAPTER 1
Conclusion
AIS4YWs turns big frameworks into small, durable habits. With GymComp, youth workers practise verify → create, co-govern participation, and privacy by default until these moves become routine; with the Workshops, they package what works into a multilingual Learning Object other practitioners can adopt the next day. Together, these strands link DigComp 2.2 and LifeComp to everyday facilitation, keeping ethics, safety and inclusion inside the work rather than bolted on. This approach squarely serves the Erasmus+ priorities—quality and innovation in youth work and inclusion & diversity—by building a shared, cross-border method rather than a tool list. Roles are clear (Arcipelago coordinates GymComp and dissemination; Esplai leads monitoring, impact, and the LO) and the timeline embeds learning, evidence, and dissemination events that carry results beyond the cohort.
The project’s value also travels further: sessions and assets plug into existing youth programmes, civic education, libraries and community centres, while partner networks in Italy and Spain amplify reach and sustain practice after funding. In short, AIS4YWs offers a practical bridge from European policy to street-level education: repeatable circuits, lightweight evidence, and open resources that help young people create, verify, participate—and protect themselves and others—using AI responsibly If the sector holds to three commitments—verify then create, privacy by default, and co-governed participation—youth workers don’t need to be AI engineers to lead meaningful learning. They need a humane method, shared assets, and peers across Europe. AIS4YWs contributes all three.
6) Safety & privacy practices learners actually retain
Privacy sessions move beyond definitions into routine behaviours: unique passwords, managers, rotation; recognising metadata collection; reading consent flows; and understanding differences between messaging apps (end-to-end encryption defaults, backups, and metadata). Learners practise configuring settings and discuss trade-offs using concrete app examples. The legislative snapshot (GDPR, ePrivacy, Data Governance Act) is simplified into three learner promises: “collect less”, “share consciously”, “retain briefly”, explained through everyday cases such as cookies and algorithmic recommendations in social platforms.
4. Competence development through GymComp
The GymComp methodology, structured around DigComp 2.1 and LIFEComp, played a key role in enabling self-assessment and capacity-building: Participants reflected on DigComp areas such as 1.1 Browsing and Searching, 2.3 Sharing and Collaborating, 3.1 Digital Content Creation, and 4.2 Protecting Personal Data, mapping their own confidence levels before and after the workshops. In line with LIFEComp, competences such as agency, emotional awareness, and learning-to-learn were embedded through role-play and scenario building. Several youth workers reported that this framework not only supported their own growth, but also gave them a language to talk about digital and transversal competences with the young people they support. As a result, over 75% of pilot participants stated they had developed new skills that would support AI adoption in their professional practice. Moreover, 84% indicated they would recommend the resources to peers, confirming both the pedagogical value and transferability of the approach.
1. Opportunities: personalization, creativity, and inclusion
Workshop participants across Spain and Italy emphasized the potential of AI to personalize educational content, foster creativity, and open new avenues for inclusion: Generative tools such as ChatGPT enabled youth workers to design fictional personas and learning scenarios tailored to real-world contexts. For example, participants created personas such as Amina, a young legal assistant, for whom they developed a full learning path supported by AI-generated content. Brisk and MagicSchool were used to quickly generate quizzes, interactive exercises, and adapted materials. These tools were especially praised for helping educators simplify complex texts or create differentiated resources according to learners’ linguistic and cognitive needs. The impact matrices completed in the workshops highlighted that AI can facilitate access for young people with disabilities or language barriers—aligning with the LIFEComp dimensions of inclusiveness, learning-to-learn, and openness to diversity. This echoes the AI4YouW pilot results, where 84% of youth workers found the resources applicable to real-life youth work, and 78% reported improved ability to use AI in professional settings.
Objectives
Activities
Designing an AI-Enhanced Workshop
1. Opportunities: personalization, creativity, and inclusion
Workshop participants across Spain and Italy emphasized the potential of AI to personalize educational content, foster creativity, and open new avenues for inclusion: Generative tools such as ChatGPT enabled youth workers to design fictional personas and learning scenarios tailored to real-world contexts. For example, participants created personas such as Amina, a young legal assistant, for whom they developed a full learning path supported by AI-generated content. Brisk and MagicSchool were used to quickly generate quizzes, interactive exercises, and adapted materials. These tools were especially praised for helping educators simplify complex texts or create differentiated resources according to learners’ linguistic and cognitive needs. The impact matrices completed in the workshops highlighted that AI can facilitate access for young people with disabilities or language barriers—aligning with the LIFEComp dimensions of inclusiveness, learning-to-learn, and openness to diversity. This echoes the AI4YouW pilot results, where 84% of youth workers found the resources applicable to real-life youth work, and 78% reported improved ability to use AI in professional settings.
10) Why the LifeComp layer matters in AI practice
Beyond technique, Gym Comp relies on empathy and facilitation to reconnect young people, communities and institutions. Materials from the Italian outreach tradition formalise empathy as a methodological tool—suspending judgement, listening actively, and co-designing steps out of discomfort—so AI-enhanced activities remain human-centred and restorative. Bottom line: Gym Comp blends DigComp’s “can do” with LifeComp’s “how we do it together”. The materials translate EU guidance and partner experience into repeatable circuits where youth create, verify, participate—and protect themselves and others—using AI responsibly.
CHAPTER 3
Reflexions
Gym Comp works because it treats EU frameworks as means, not ends. The materials start from real youth-work frictions—privacy fatigue, tool access limits, “AI knows everything” myths—and translate DigComp/LifeComp into repeatable routines that learners can actually do under time pressure. That is the decisive shift: competences are not lists on a wall but behaviours embedded in short circuits—check the source, name the risk, set the control, reflect as a team. This framing mirrors LifeComp’s awareness–understanding–action arc and makes the “relational layer” visible rather than ornamental. A second takeaway is the centrality of safety-by-default. When privacy and security appear only as an add-on, they lose to convenience; when they are baked into every task (cookies, consent, passwords, messaging settings), learners start to narrate their own privacy stories and make concrete changes (e.g., enabling MFA, rotating passwords, choosing E2E defaults). The Italian privacy deck helps by turning abstract rules (lawful basis, purpose limitation, minimisation, storage limits) into short, memorizable prompts that participants can reuse across platforms. In practice, teaching the why of GDPR without the how of settings and habits is ineffective; Gym Comp keeps them coupled.. The sessions also surface an equity issue often ignored in AI trainings: access and licensing. Partners document how free tiers shrink and upload limits change, pushing learners toward paid plans. That matters in youth work, where the promise of “AI for everyone” collapses if activities depend on gated features. The honest response is not to lament but to design tool-agnostic exercises (verify, attribute, protect) and plan for institutional accounts or open alternatives when a licensed tool is pedagogically necessary. This keeps inclusion non-negotiable, not aspirational.
On the ethics and trust front, the ALTAI lens proves practical when collapsed into a few recurring questions: Who is affected? Where are the data from? Where is the human in the loop? What can go wrong, and how will we notice? In workshops, these prompts helped move from “AI is cool/scary” to auditable practice—traceable outputs, explicit attribution, and an agreed human review step. Rather than policing language, the checklists scaffold judgement and allow youth workers to justify choices to peers, funders, and participants.Still, AI risk literacy must be taught as technique and mindset. Slides on “modern AI hype” make a useful distinction: models can dazzle on unstructured data and still hallucinate confidently; learners need both prompting discipline and external grounding to resist that pull. In our view, the most robust habit taught here is “verify first, then create”: start from sources, constrain the model with them, and cite. This counters magical thinking without dampening creativity. The civic participation strand is where facilitation is most tested. Spanish discussions show how participatory platforms can drift into popularity contests and rivalry without guardrails. That is not a reason to abandon them; it is an argument for explicit norms, transparent data use, and bias-aware moderation designed with young people, not for them. Gym Comp’s choice to prototype participation rules inside the session is the right move: it treats youth as co-governors of the digital spaces they use.
Objectives
Activities
CHAPTER 3
Reflexions
In recent years, the debate on Artificial Intelligence regulation has become central. AI is not a neutral technology: its decisions and uses can profoundly affect people’s lives, the market, and society. Why is regulation needed?
- Protection of fundamental rights – AI manages large amounts of personal data and can affect fundamental freedoms such as privacy, equality, and non-discrimination.
- Legal responsibility – Who is accountable if an AI system causes harm? It is necessary to define clear chains of responsibility, including developers, providers, and users.
- Promotion of trust and innovation – Clear regulation increases citizens’ trust and encourages safer investments by companies.
Ethical and social dimension of AI for critical and conscious use- AI must be anthropocentric, serving human well-being and the common good.
- It must support democratic processes, fundamental rights, and the rule of law, avoiding risks such as mass surveillance or discriminatory bias.
Continuous and participatory reflection between developers, policymakers, and civil society is necessary to address ethical dilemmas and tensions between principles (e.g., security vs. individual freedom).CHAPTER 1
Conclusion
The integration of AI into youth work is not simply a matter of adopting new tools—it challenges the very foundations of how we understand learning, agency, and the human relationship in education. While the AI4YW project and its predecessor demonstrated that AI can expand access, improve personalization, and foster engagement, they also revealed significant tensions and blind spots that require ongoing attention. Rebalancing power and agency One of the central questions emerging from the workshops is: Who controls the learning process when AI is involved? While generative tools empower users to create, simulate, and personalize, they also centralize technological power in the hands of opaque systems, often operated by private actors. Youth workers highlighted a need to preserve learner agency by ensuring AI is used as a scaffold—not a substitute—for critical thought, creativity, and collaboration. The GymComp approach, rooted in LIFEComp’s emphasis on personal and social competence, helped bring this into focus. By combining technical upskilling with transversal reflections, participants began to reframe AI not as a tool for efficiency alone, but as a space to question, negotiate, and co-create meaning. Moving beyond the "Toolbox" mentality A recurring insight was the importance of moving beyond a purely instrumental view of AI. The tendency to treat AI as a collection of apps risks fragmenting its ethical, social, and political implications. Instead, what emerged was a call to situate AI within the lived realities of youth—where issues of identity, inclusion, bias, and trust are constantly at stake.
Youth work, especially in non-formal education, is inherently relational. The best moments of the AI4YW process were not about what AI could do on behalf of educators, but about what it could enable in dialogue with them. These insights echo broader calls from the youth sector for a critical pedagogy of technology: one that puts human values, collective intelligence, and democratic oversight at its core.From Digital Competence to Digital MaturityDigComp 2.1 was essential in structuring the workshops, but the project experience showed that digital competence alone is not enough. What youth workers need is a form of digital maturity: the ability to make informed choices, critically evaluate systems, and support young people in navigating a complex technological landscape. This maturity cannot be acquired solely through checklists or tutorials—it requires time, dialogue, and reflective practice.In this sense, the GymComp model offered a useful prototype for how digital and transversal competences can be cultivated holistically. It recognized that understanding AI is not just about prompt engineering or data literacy—it’s about developing a mindset that combines ethical awareness, emotional intelligence, and adaptive learning. Toward a human-centered AI culture in Youth Work Finally, the project affirmed that integrating AI into youth work is not only about training or tools—it is about culture. The culture of youth work values participation, inclusion, care, and empowerment. For AI to be genuinely transformative, it must be subordinate to these values, not the other way around.
3. Ethical reflection and responsible use
Ethics were not treated as an abstract add-on, but as a central pillar of all workshop activities. Drawing on the UNESCO framework and the LIFEComp focus on responsibility, collaboration, and critical thinking, the AI4YW project introduced a range of reflective tools: Participants engaged with examples such as the MIT “Moral Machine” experiment, discussing how AI might reinforce or challenge social biases. In collaborative settings, they identified situations in youth work where AI might undermine equity (e.g., automated profiling in job training programs), and co-created guidelines for responsible use. The importance of human oversight and the non-substitutive role of AI was stressed. AI was seen as a co-pilot—not a replacement—for the empathy, ethics, and relational work of youth professionals. This aligns with findings from AI4YW the pilot, where 78% of respondents affirmed that the OERs helped them recognize AI bias and promoted ethical awareness.
CHAPTER 3
Conclusions
Building ethical AI requires multidisciplinary collaboration between:
- Developers and engineers, who understand the logic of models;
- Lawyers and legislators, who define the regulatory framework;
- Sociologists and ethicists, who assess social and cultural impact;
- End users, who provide real feedback on the use of technologies.
Ethics does not limit innovation: it makes it reliable and socially acceptable, creating trust and 'responsible competitiveness.' The ethical aspect must be incorporated into every phase of the AI system life cycle according to the principle of 'Ethics by Design.DigiCompt 3.1. - Digital content creation