Skip to main content

Children and Adolescents in the Age of Generative AI: 5 Reactions and prevention

Children and Adolescents in the Age of Generative AI
5 Reactions and prevention
  • Show the following:

    Annotations
    Resources
  • Adjust appearance:

    Font
    Font style
    Color Scheme
    Light
    Dark
    Annotation contrast
    Low
    High
    Margins
  • Search within:
    • My Notes + Comments
    • Notifications
    • Privacy
  • Publication HomeChildren and Adolescents in the Age of Generative AI
  • Publications
  • Learn more about Manifold

Notes

table of contents
  1. Cover image
  2. Title Page
  3. Imprint
  4. Logos
  5. Contents
  6. Figures
  7. Tables
  8. Executive Summary
  9. 1 Introduction
  10. 2 AI as disruptive technology
  11. 3 AI in the school context
  12. 4 The situation in Luxembourg
  13. 5 Reactions and prevention
  14. References
  15. Glossary of Key Terms

5 Reactions and prevention

The following chapter evaluates the services offered by BEE SECURE with regard to the integration of AI topics. The focus is on questions of coverage as well as aspects of educational communication. The chapter also deals with the ethical challenges arising from the work of a Safer Internet Centre. Finally, suggestions are presented on how the work of BEE SECURE can be adapted to the challenges of AI.

5.1 Analysis of existing BEE SECURE offerings

BEE SECURE, the Luxembourgish Safer Internet Centre, is part of a pan-European initiative co-funded by the European Commission under the Digital Europe Programme. Safer Internet Centres (SICs) operate across EU Member States to promote a secure and empowering digital environment for children and young people. Each centre typically combines three core functions: an awareness node that develops educational campaigns and resources on online safety; a helpline that offers confidential advice on issues such as cyberbullying, privacy, and digital well-being; and a hotline for reporting illegal online content, including child sexual abuse material. Beyond these services, SICs foster youth participation through advisory panels and collaborate internationally via the Better Internet for Kids network. Their work reflects a broader European strategy to enhance digital literacy, mitigate online risks, and uphold fundamental rights in the digital sphere. Internal survey data among BEE SECURE staff indicate that trainers and advisors feel generally well prepared to address AI-related questions across training sessions, awareness activities, and Helpline/Stopline support, which further strengthens the foundation of the current service portfolio.

BEE SECURE’s activities are extensive: it delivers over 1,200 training sessions annually in schools, youth centres, and community spaces, covering topics from cyberbullying and sextortion to data privacy and misinformation. A network of 15 freelance trainers plays a key role in identifying emerging issues and engaging directly with students through youth panels, ensuring that BEE SECURE remains attuned to the latest digital behaviours. Beyond training, BEE SECURE organises around 40 events per year, including Safer Internet Day and thematic workshops, and regularly launches national campaigns. One notable example is the Cyberjungle campaign, which uses playful storytelling and AI-generated content to raise awareness about deepfakes, sextortion, and romance scams under the slogan “Gleef net alles am Netz” (“Don’t believe everything online”). These campaigns are informed by Hot Topics meetings, in which BEE SECURE collaborates with stakeholders to define priorities.

A distinctive feature of BEE SECURE is its Radar report, published annually to monitor trends in young people’s digital habits. Based on surveys of thousands of students, parents, and educators, the Radar provides insights into issues such as screen time, social media use, cyberbullying, and perceptions of AI. For example, the 2025 edition revealed that Snapchat and WhatsApp dominate among teens, while concerns about sextortion and misinformation are rising sharply. The Radar serves as an evidence-based tool for policymakers, educators, and parents, guiding national strategies like the sécher.digital action plan. Alongside this, BEE SECURE publishes guidelines, news, and educational materials tailored to different audiences, reinforcing its role as a central resource for digital literacy and online safety in Luxembourg.

BEE SECURE pursues a multidimensional approach to the prevention and intervention of digital risks. This includes the services Helpline and Stopline, both based at the Kanner-Jugendtelefon (KJT), a wide range of training and awareness-raising activities for children and young people, as well as events for parents and educational professionals. In addition, BEE SECURE regularly publishes information materials and guidance documents aimed at promoting media literacy and digital safety.

5.1.1 Stopline and Helpline

The Stopline serves as Luxembourg’s national reporting hotline for illegal online content and is operated by Kanner-Jugendtelefon (KJT) with national BEE SECURE partnership. It is a member of the International Association of Internet Hotlines (INHOPE) and operates in line with INHOPE’s binding Code of Practice. The Stopline receives, assesses and forwards reports of potentially illegal online material to the Police Grand-Ducale and, through the INHOPE network and the ICCAM platform, to international partners such as INTERPOL. Its primary focus lies on identifying and facilitating the removal of child sexual abuse material (CSAM), including an increasing number of synthetically generated depictions and deepfakes involving minors.

The Helpline provides confidential counselling for children, young people, parents and teachers on all questions relating to safe and responsible internet use. Counselling is available by telephone, online chat and email. While telephone calls are anonymous, written forms of contact (chat or email) are confidential but not anonymous. The Helpline is part of the European Insafe network, which is coordinated by the European Commission under the Better Internet for Kids programme, and provides guidance and emotional support in case of Cyberbullying, data protection concerns, and other online risks.

Both services make a vital contribution to strengthening the digital resilience and safety of young people in Luxembourg. They operate in accordance with European quality standards, are internationally connected (INHOPE, Insafe), and firmly embedded within the national child-protection framework.

In recent years, the digital risk landscape has changed considerably. Alongside traditional CSAM, there has been a marked increase in AI-generated abuse material (“synthetic CSAM”) and deepfakes depicting real children or adolescents in sexualised contexts. Algorithmically amplified hate speech, disinformation and fraud schemes involving AI components — such as fake identities or chatbots — are also on the rise. In addition to established topics such as cyberbullying, data protection, excessive media use and risky online challenges, there has been an increase in deepfake-based cyberbullying cases, nudifying apps, and AI tools that generate intimate forgeries leading to sextortion or digital blackmail.

Although systematic statistical data collection on AI-related reports does not yet exist, Helpline and Stopline staff confirm the growing relevance of such cases in their daily counselling and reporting practice. Establishing a structured monitoring and statistical analysis of AI-related phenomena would therefore be advisable to identify trends and target preventive action more effectively. Despite the absence of systematic case statistics, the internal staff survey shows that Helpline and Stopline advisors generally feel confident in handling AI-related incidents and are able to draw on established procedures and professional routines when supporting affected users.

Within the framework of international hotline cooperation – particularly within the INHOPE network, of which the Luxembourg Stopline is an active member – several key themes and reporting categories can be identified that have gained new momentum through the use of AI. According to the INHOPE Annual Report (2024), three main developments can be observed across member hotlines:

  1. A significant increase in deepfake-related CSAM, including synthetically generated depictions of minors;
  2. A shift of illegal content to encrypted or difficult-to-monitor platforms (e.g. Telegram);
  3. A growing connection between AI tools and online offences, as easily accessible image generators and chatbots are increasingly being used to prepare or conceal criminal acts.

More generally, AI technologies are also being used to conceal identities, manipulate images or text, and create fraudulent digital profiles or phishing schemes that mimic real individuals. While not all such cases are criminally relevant, they represent a new layer of technical and ethical complexity in hotline and counselling work, requiring continued monitoring and specialised staff training. According to the internal staff survey, these AI-mediated forms of sexualised harm are perceived by advisors as particularly demanding and emotionally challenging, often requiring extensive guidance, reassurance, and safety planning during Helpline interactions.

Through regular participation in national and international training programmes, supervision and network exchanges, staff members ensure high professional standards and maintain up-to-date expertise in addressing emerging digital phenomena, including synthetic CSAM, deepfake-based harassment, and AI-supported online manipulation. According to the internal staff survey, these training opportunities are perceived as highly effective and contribute significantly to advisors’ sense of confidence when dealing with AI-related cases in their daily practice. The continuous professional development safeguards the consistent application of established procedures and ensures that Luxembourg’s Stopline and Helpline services continue to play a leading role in child online protection.

5.1.2 Publications

BEE SECURE’s publications are freely accessible online, providing an easily available and low-threshold way to obtain information about digital risks and protective measures.

Table 3: BEE SECURE publications: AI relevance and suggested content updates.
No.Publication AI relationRecommended additions/updates (keywords)
1BEE SECURE Radar 2025Yes – analyses AI awareness, deepfakes and the AI Act.Multimodal generators, voice cloning, latest legislation.
2Rapport d’activité 2024Yes – reports on AI training and deepfake campaigns.Evaluation of the campaigns, AI Act links, AI voice cloning.
3Thematischer Beitrag „KI – Chancen & Risiken“Yes – explains AI, machine learning and the AI Act.Generative models, voice cloning, use in the classroom.
4Poster „Passwörter der Schüler verwalten“No – password tips.Passkeys, two factor authentication, AI password cracking risks.
5Thematischer Beitrag „Multi Level Marketing & Co.“No – explains MLM.Warning about AI trading bots, deepfake advertising, AI bots.
6Thematischer Beitrag „Hate Speech“No – defines hate speech.AI generated hate speech, algorithmic moderation.
7Thematischer Beitrag „Dark Patterns“No – UX manipulation.AI based personalised dark patterns, regulation.
8Ratgeber „Bist Du Opfer von Cyber Mobbing?“No – explains bullying and the law.Deepfake bullying, voice cloning, AI bots as perpetrators.
9Thematischer Beitrag „Influencer“Limited – mentions virtual influencers.AI influencers, deepfake advertising, recommendation systems.
10UE „Sexting“No – sexting risks.Deepfake sextortion, AI bots, labelling requirement (AI Act).
11Cyber Mobbing KitNo – teaching units on bullying.Deepfake bullying, AI moderation and detection.
12PEGI Label für VideospieleNo – explains age ratings.Generative AI in games, AI age verification, loot box algorithms.
13Ratgeber „Nackt im Netz?“No – sexting/privacy.Deepfake pornography, AI image search, protection against AI manipulation.
14UE „Selbstdarstellung – Mein perfektes Ich“No – reflects on virtual vs. real self.AI filters, recommendation systems, deepfake effects.
15UE „Selbstdarstellung – Wie authentisch sind Influencer?“No – examines influencer culture.Role of algorithms, generative content tools, deepfakes.
16UE „Sharenting – Familienfotos im Internet“No – discusses privacy.AI facial recognition, data collection, deepfake risks.
17UE „Sexting – For Your Eyes Only“No – rules for sexting.Deepfake sextortion, AI extortion bots, protective tools.
18DigiRallye – Virtual EditionIndirect – station “The Internet lies” touches on deepfakes.Dedicated station on AI/deepfakes, voice assistants, recommendation systems.
19CYF Dossier „Desinformation in den Medien“No – journalistic fact check.AI fake news, generative text/images, detection tools.
20Elternratgeber „Bildschirme in der Familie“No – screen time guide.Recommendation algorithms, voice assistants, deepfakes, child protection filters.
21Dossier „Desinformation in der Politik“Indirect – explains algorithms, social bots, micro targeting.Generative AI in political disinformation, deepfakes, AI Act notes.
22Elternratgeber „Mein Kind im Internet“No – general online tips.AI recommendation algorithms, voice assistants, deepfake filters.
23Data Detox KitYes – warns against deepfakes and “cheap fakes.”Modern generative models, voice cloning, deepfake detection.
24„Was tun bei einer Datenschutzverletzung?“No – steps after a data breach.AI phishing, AI based fraud detection, current legislation.
25Dossier „Desinformation“No – warns about algorithms and filter bubbles.Generative AI fake news, labelling synthetic media, AI Act.
26Internet Sicherheitskonzept für Maisons RelaisNo – security concept for childcare.AI assistant systems, deepfake prevention, algorithmic filters, AI Act.
27Flyer „Cybermobbing: Anzeige erstatten?“No – legal advice.Deepfake bullying, voice cloning, tools to detect AI generated harassment.
28Checkliste „Vernetzte Spielzeuge“No – smart toy safety checklist.AI features (voice/facial recognition), data protection, legal requirements.
29Thematischer Beitrag „Algorithmen – Big Data“Yes – explains AI, neural networks, deep learning, the Turing Test and machine learning.Generative AI (ChatGPT, diffusion models), ethics/bias, AI Act, current examples.
30Thematischer Beitrag „Big Data: Datensammlung im Alltag“Indirect – mentions AI use in data analysis and Cambridge Analytica.Current AI analytics, generative models, data ethics, AI Act, predictive analytics.
31NetiquetteNo – rules of conduct and politeness.Rules for dealing with AI chatbots, deepfake warnings, recognising generated content.
32„Auch digital ein Vorbild sein“No – guide for educators/teachers on online reputation and privacy.Explain algorithmic recommendation systems, deepfake awareness, use of generative tools in teaching.
33„Hate Speech und das Gesetz“No – legal framework for hate speech.AI generated hate speech, detection algorithms, AI Act legal obligations.
34Dossier „Social Bots“Yes – explains bots and emphasises rapid progress in AI.Generative chatbots (LLMs), voice/video bots, transparency obligations (AI Act).
35Dossier „Filterblasen und Echokammern“Indirect – describes algorithmic personalisation.AI based recommendation systems, generative content, transparency and regulation (AI Act).
36Thematischer Beitrag „Recht am eigenen Bild“No – legal basics on photos and privacy.Deepfakes and AI editing, automated facial recognition, tools for detecting doctored images.
37E Booking: Betrug beim Online BookingNo – explains booking fraud.Dynamic pricing, AI generated fake reviews, AI phishing emails.
38E Shopping: Betrug beim Online ShoppingNo – explains fake shops, escrow fraud.AI generated fake shops, fraudulent chatbots, deepfake product reviews.
39E Banking: Betrug beim Online BankingNo – explains phishing, trojans.AI phishing, AI fraud detection, biometric authentication.
40Thematischer Beitrag „Pornografie“No – describes pornography and youth protection.Deepfake pornography, AI content filters, legal classification of AI pornography.
41Dossier „Darknet – Die dunkle Seite des Netzes“No – explains Tor and illegal vs. legal use.AI supported surveillance, automated malware generation, deepfake marketplaces.
42Gegenrede – Strategien gegen Hate SpeechNo – tips for counter speech.AI amplified hate speech, automatic moderation, AI based counter speech bots.

The analysis of 42 BEE SECURE publications reveals a broad thematic spectrum ranging from digital education, data protection and fraud prevention to hate speech and pornography. Only a minority of the examined documents explicitly address artificial intelligence or machine learning: BEE SECURE Radar 2025 analyses current levels of knowledge about AI, outlines risks such as deepfakes, and explains regulatory initiatives such as the AI Act. The Rapport d’activité 2024 reports on training sessions concerning AI and deepfake campaigns, while the thematic publication “Artificial Intelligence – Opportunities and Risks” introduces key concepts, applications, and potential dangers of AI. Also linked to AI are the dossiers “Social Bots”, which highlights the increasing sophistication of such bots due to advances in artificial intelligence, and “Algorithms – Big Data”, which explains neural networks, deep learning and the Turing Test.

The majority of publications, however, focus on traditional aspects of digital safety, such as fraud in online shopping and banking, privacy protection, dealing with hate speech, or the legal framework for photography. In these documents, artificial intelligence is either not mentioned at all or is only indirectly referenced through the term “algorithms”. This points to a gap between the rapid evolution of AI-based systems (generative AI, deepfakes, chatbots) and the current informational basis.

It would therefore be advisable to expand the existing publications to include aspects related to AI-driven threats and emerging fraud schemes, such as voice cloning or AI phishing. In addition, AI-based moderation and detection tools, as well as current legal frameworks such as the AI Act, should be taken into account. Particular attention should also be given to issues such as deepfake pornography, personalised dark patterns, algorithmic recommendation systems and AI-supported hate speech filters, in order to adapt the publications more effectively to the realities of the digital age. To further enhance the accessibility and relevance of its professional learning resources, it would also be advisable for BEE SECURE to include clear publication dates and version information on all materials and future updates. This would allow educators and institutions to verify at a glance whether they are consulting the most recent version or whether newer editions are available, thereby ensuring that guidance remains accurate and up to date in a rapidly evolving digital environment.

5.1.3 Training sessions for school classes

Table 4: BEE SECURE trainings for pupils and students: pedagogical methods, AI relevance, suggested content updates.
TrainingTarget group & durationPedagogical methodAI relationNeed for update
Ein Sprung ins NetzCycle 3, 90 minutesThe class undertakes a fictional “walk” through familiar everyday situations which are then mirrored by online scenarios. Through games and discussions they explore passwords, codes of conduct, contact with strangers and cyberbullying.No AI topics are covered.The content could be expanded to include warnings about AI driven phishing methods, password cracking algorithms or personalised dark patterns.
Cyber Mobbing: Mit Detektiv Shadow auf MissionCycle 3.2, 90 minutesChildren work with the fictional character “Detektiv Shadow” to solve puzzles and uncover the issue of cyberbullying. They then create a “cyberbullying expert kit” with tips for dealing with such incidents.No AI content; the focus is on netiquette, image rights and the BEE SECURE helpline.Additional topics such as deepfakes and voice cloning as new forms of bullying, and AI based detection tools, could be addressed.
BEE FRIENDS – Alex’ erste Schritte onlineCycle 4, 90 minutesStep by step, the class creates the social media profile of a fictional character (“Alex”) and jointly decides how to react to typical situations. The decisions and BEE SECURE key messages are recorded on a poster.No explicit AI content; topics include passwords, self presentation, privacy, competitions and chain letters.It would be advisable to explain algorithmic recommendation mechanisms on social networks, as well as generative filters and chatbots.
Hinter den Kulissen der sozialen NetzwerkeCycle 4.2, 90 minutes (only bookable after the basic module “BEE FRIENDS”)A change of perspective: pupils take on the role of app developers and playfully consider risks, data collection and advertising models. They learn about protective measures and are encouraged to question influencers.No direct AI content; the focus is on rules of conduct, inappropriate content, advertising, influencers and data collection.AI driven personalised advertising, deepfake influencers and recommendation algorithms could be incorporated.
Fit fürs Netz?Secondary school, Year 7e, 90 minutes (part of the “Digital Sciences” curriculum)Using the virtual desk of a character (“Dennis”), young people work through five topics. Each object on the desk leads to a new case study on passwords, emails, sexting, self presentation and big data.Big data aspects are addressed; AI is not mentioned explicitly.The link between big data and AI (e.g. profiling, recommendation algorithms) and current AI risks such as deepfake sexting could be added.
Hype im Netz – so bleibst du sicher und informiert!Secondary school, Years 7e and 6e, 90 minutesStudents again explore “Dennis’s desk”, this time focusing on very current topics. Case studies on deepfakes, disinformation, hate speech, cybergrooming and online challenges stimulate discussion .Explicit AI topics are included: deepfakes and artificial intelligence are presented as part of the risks.The contents should be updated regularly to reflect new AI trends (e.g. voice and video generators) and legal developments such as the AI Act.
Fact Check – Safe and SoundSecondary school, Years 6e and 5e, 90 minutesA role play: young people become festival organisers and must fend off phishing attacks, fake news campaigns and disinformation. They use fact checking tools (e.g. reverse image search) and learn to recognise filter bubbles and algorithms.AI is part of the training; deepfakes and algorithms are addressed, and the use of AI in disinformation is explained.Continuous updates to include new deepfake technologies, AI driven disinformation campaigns and emerging regulations are recommended.
Lost in Data – Wo werden im Alltag Daten gesammelt?Secondary school, from Year 4e upwards, 90 minutesStorybased learning: the class accompanies a fictional couple to Barcelona and searches for the missing Nino via digital traces. Students learn where digital footprints arise in everyday life and how big data, algorithms and AI affect their privacy.This training has a clear AI component; it explains how algorithms and AI are used in data collection and raises awareness of big data analytics.Updates could take account of newer forms of data analysis (e.g. generative AI and personalised advertising systems) and refer to forthcoming regulations such as the AI Act.

BEE SECURE makes an important contribution to prevention through its training programmes for school classes. While the basic modules in primary education (“Ein Sprung ins Netz”, “BEE FRIENDS”, “Superhelden im Einsatz gegen Cybermobbing”) focus primarily on topics such as passwords, privacy and respectful online behaviour, aspects related to artificial intelligence have so far received little attention. In view of the rapid development of generative models and the easy accessibility of AI tools, these modules should in future be expanded to include age-appropriate explanations of algorithmic recommendation systems, deepfake filters and privacy-friendly default settings. A promising approach lies in linking familiar offline situations with corresponding online risks — a method already used by BEE SECURE in “Ein Sprung ins Netz”. This approach can also help illustrate new risk scenarios, such as chatbots appearing in games or personalised advertising.

At the secondary level, BEE SECURE trainings demonstrate how AI-related topics can be taught in a concrete and practice-oriented way. “Lost in Data” takes students on a digital treasure hunt through Barcelona, allowing them to experience big data analysis, algorithms and AI in data collection first-hand. “Fact Check – Safe and Sound” engages participants in organising a music festival while confronting them with deepfakes, phishing attempts and filter bubbles — thereby equipping them with essential tools such as reverse image search and fact-checking platforms. The workshop “Hype im Netz” directly addresses current trends such as deepfakes, cyber grooming and artificial intelligence, showing how quickly online hypes spread and what mechanisms drive them. The hands-on nature of these workshops, which use real tools and problem-solving role plays, strengthens critical media literacy and underlines that AI is not an abstract concept but an integral part of young people’s everyday lives.

For a future-oriented media education programme, this White Paper recommends systematically embedding AI-related topics across all age groups. This includes early awareness of AI-driven manipulations such as voice cloning, personalised dark patterns and recommendation algorithms. At the same time, learners should develop practical skills — including the use of detection tools for deepfakes, questioning algorithmic decisions, and reflecting on the data they voluntarily disclose.

5.1.4 Trainings for non-formal education groups

Table 5: BEE SECURE non-formal trainings: pedagogical focus, AI relevance, and suggested content updates.
TrainingTarget group & durationPedagogical methodAI relationRecommended additions/updates (keywords)
Die Abenteuer von Captain Kara7–9 years, 2 hoursChildren watch a Childnet International cartoon featuring Captain Kara and the SMART Crew. In small groups they discuss case studies from the film and work out safe solutions for email scams, fake profiles and other online risks.No AI topics; focus is on basic internet safety (e mails, disinformation, privacy, cyberbullying, fake profiles).Introduce examples of AI driven scams such as chatbots, generative fake profiles or AI powered phishing emails; explain how deepfakes and synthetic voices could make disinformation more convincing.
Superhelden im Einsatz gegen Cyber Mobbing (extracurricular)7–9 years, 2 hoursWith the mascot Panda Wanda, children tackle adventures and quizzes that teach them to recognise and stand up to cyberbullying. Completing the tasks earns them a superhero ID card encouraging supportive behaviour.No AI content; the focus is on cyberbullying, netiquette and where to seek help.Expand to include new forms of bullying involving AI, such as deepfake manipulation or voice cloning, and introduce simple detection tools.
Quiz dich fit!8–12 years, 2 hoursDetails are not provided, but the title suggests a quiz based format where children test and expand their knowledge of online safety.No explicit AI topics.Ensure that quiz questions cover emerging AI driven threats such as AI chatbots, recommendation algorithms and fake content to raise awareness in a playful way.
Let’s talk about Selbstdarstellung im Netz13–16 years, 90 minutesA non formal discussion workshop about self presentation on YouTube, TikTok and Instagram. Participants and educators exchange experiences and opinions using games, videos and other methods, emphasising privacy and data awareness.No explicit AI topics; focuses on self presentation, privacy, digital consumption and data collection.Incorporate discussion of AI driven recommendation systems, generative filters and algorithmic impacts on self esteem; highlight how influencers may use AI tools for content creation.
Let’s talk about Sexting13–16 years, 90 minutesA moderated conversation about sexting in a non formal setting. Using games and videos, young people reflect on their experiences, learn about consent and privacy, and discuss sextortion and image rights.Likely no AI topics; emphasis on sexting, sextortion, privacy and image rights.Add information on AI generated deepfake pornography, sextortion bots and tools to verify authenticity of images; discuss the role of AI in amplifying risks and the importance of critical awareness.

Beyond the classroom, BEE SECURE also addresses children and young people in non-formal education settings. The programmes for 7‑ to 9‑year-olds take a strongly narrative approach: “Die Abenteuer von Captain Kara” uses an animated film and the SMART Crew to highlight the dangers of scam emails, disinformation, cyberbullying and fake profiles, while “Superhelden im Einsatz gegen CyberMobbing” invites participants to join Panda Wanda’s “superhero club” by mastering adventures and quizzes about respectful online behaviour. A quiz-based workshop, “Quiz dich fit!”, aims to reinforce basic knowledge about internet safety in a playful way for 8- to 12-olds.

Older adolescents are offered discursive spaces rather than readymade lessons. In “Let’s talk about Selbstdarstellung im Netz”,‑ 13‑ to 16‑year-olds reflect on how they present themselves on Instagram, YouTube or TikTok, examine the balance between authenticity and staging, and discuss privacy and data consumption habits. The companion session “Let’s talk about Sexting” tackles the sensitive issues of sexting, sextortion and image rights, with facilitators using games and video clips to encourage open dialogue.

Although these workshops successfully build digital literacy and promote critical thinking, they currently pay little attention to the growing influence of AI on children’s online experiences.

To remain ahead of emerging threats, the storytelling and quiz formats for younger children could be updated to include AI-driven scams, such as chatbot imposters or generative fake profiles, and to explain in simple terms how deepfake images or voices are created. In the discussion-based youth sessions, it would be valuable to examine how algorithmic recommendation systems shape self-image and to raise awareness of ‑AI generated pornography and extortion techniques.

5.1.5 Trainings for teachers and educators

Table 6: BEE SECURE teacher trainings: pedagogical focus, AI coverage, and suggested updates.
TrainingTarget group & durationPedagogical methodAI relationRecommended additions/updates (keywords)
TikTok, Instagram & Co. – Was machen Kinder und Jugendliche eigentlich derzeit online?Teachers and educational staff; 3 hours; delivered in German and LuxembourgishA three hour seminar providing an overview of young people’s media habits on platforms like YouTube, Instagram, Snapchat and TikTok. Participants learn why children are drawn to specific content, explore current media trends and discuss related risks. Topics include challenges, pranks, influencers, gaming, self presentation and sexting.No explicit AI content; focus is on understanding behaviour and trends.Incorporate discussion of AI driven recommendation algorithms, the role of AI in shaping viral challenges or pranks, and how generative tools influence self presentation and influencer culture.
Sexuelle Darstellungen im Netz: Was reizt Jugendliche?Teachers and educational staff; 2.5 hours; delivered in German and LuxembourgishOnline training examining how and why adolescents encounter sexualised content on the internet. It covers Luxembourg’s legal framework and addresses risks such as sexting, sextortion and cyber grooming, with practical guidance for schools and recommended information sources.No AI content; emphasis is on legal context and safeguarding.Update with information about AI generated deepfake pornography, AI enabled sextortion bots and tools for verifying the authenticity of images or videos.
Sicherheit im Fokus: Von KI Grundlagen zur praktischen AnwendungTeachers and educational staff; 3 hours; offered in French, English, German and LuxembourgishA deeper dive into AI technologies and security. Participants receive an overview of AI’s development and integration into various tools, explore opportunities and risks, work hands on with AI tools, and analyse case studies in groups to develop safe, responsible approaches. Topics include AI, disinformation, data security and AI tools.This training explicitly covers AI, providing both theoretical background and practical exercises.Ensure the content reflects the latest AI developments (e.g. generative models, voice cloning), include guidance on the AI Act and ethical considerations, and regularly update the case studies.
Cybersecurity Essentials – was man als Lehrkraft über Internetsicherheit wissen sollteTeachers and educational staff; 3 hours; delivered in LuxembourgishInteractive session tailored to participants’ questions about internet security. It explains the underlying technologies, identifies risks and offers practical tips on topics such as cloud services, backups, account management, data protection and privacy. A second part presents case studies for group problem solving.AI is not explicitly covered; focus is on general security and privacy practices.Include emerging AI driven threats, such as AI generated phishing emails, automated malware and AI based intrusion detection; provide guidance on using AI tools responsibly in the classroom.

BEE SECURE’s professional development programme for teachers and educators plays a pivotal role in strengthening digital literacy and online safety competencies within educational institutions. The training session “TikTok, Instagram & Co. – Was machen Kinder und Jugendliche eigentlich derzeit online?” provides participants with valuable insights into young people’s online habits, motivations and risks across platforms such as YouTube, TikTok and Instagram. However, given the rapid evolution of digital ecosystems, future iterations should explicitly address the impact of AI-driven recommendation systems and generative tools shaping influencer culture and viral trends.

The online seminar “Sexuelle Darstellungen im Netz: Was reizt Jugendliche?” focuses on adolescents’ exposure to sexualised online content, legal frameworks and preventive strategies. It provides a strong foundation for safeguarding discussions but would benefit from integrating AI-related risks such as deepfake pornography, AI-assisted grooming and synthetic sextortion.

The training “Sicherheit im Fokus: Von KI-Grundlagen zur praktischen Anwendung” already represents a good-practice example by combining theoretical foundations of AI with hands-on tool exploration and group-based case analysis. Regular updates are essential to include emerging technologies (e.g. voice cloning, generative image/video systems) and ensure alignment with the European AI Act and ethical guidelines.

Finally, “Cybersecurity Essentials – was man als Lehrkraft über Internetsicherheit wissen sollte” provides a comprehensive introduction to data protection, cloud computing and privacy management, delivered through interactive exercises. While AI is not yet explicitly addressed, the course could be enhanced by including current AI-driven threats (e.g. intelligent phishing campaigns or automated malware) and by discussing the responsible integration of AI tools into teaching practice.

Overall, these training modules form a coherent framework for fostering AI-aware, ethically grounded digital education. By incorporating explicit references to AI across all topics, BEE SECURE can ensure that educators not only understand the technological landscape but also feel empowered to guide children and young people in responsibly navigating an increasingly algorithmic world.

5.1.6 Training sessions for parents

Table 7: BEE SECURE parent trainings: focus areas, AI relevance, and suggested content updates.
TrainingTarget group & durationPedagogical methodAI relationRecommended additions/updates (keywords)
Kinder und Jugendliche im Internet – Eine Herausforderung für ElternParents; 2 hours; available in French, English, German and LuxembourgishBased on the BEE SECURE guide Bildschirme in der Familie, this parent evening provides practical advice on balancing opportunities and risks of children’s internet use. Topics include screen time, smartphone use, online abuse and cyberbullying.No explicit AI content; focus is on general online safety and parental guidance.Integrate information on AI-driven recommendation systems, generative content, and risks such as deepfakes or chatbot grooming; include guidance on talking to children about AI tools.
Kindersicherung im Internet: praktische Tools, Techniken und TippsParents; 90 minutes; available in LuxembourgishConducted as an online webinar, this session demonstrates settings, programmes and technical methods that enhance children’s online safety, with a practical focus on device configuration and parental-control software.No AI content; focus on technical parental-control measures.Update to include AI-based parental-control and monitoring tools, explain how algorithms recommend or restrict content, and address ethical and privacy implications.
Expertentrio: Experten beantworten Ihre FragenParents; 2 hours; available in LuxembourgishA Q&A event organised by BEE SECURE in cooperation with KJT and the Luxembourg Police’s crime prevention unit. It covers a wide range of safety issues concerning children’s internet use, including legal and social aspects.No direct AI content; focus is on prevention and expert advice.Add discussion of AI-related threats (deepfakes, AI-generated grooming, fake profiles) and guidance on how parents can recognise and report AI-driven manipulation or fraud.

BEE SECURE’s training offer for parents is designed to help families navigate the challenges of raising children in an increasingly digital environment. The sessions focus on practical guidance, open dialogue and the everyday realities of online life at home. “Kinder und Jugendliche im Internet – Eine Herausforderung für Eltern” introduces key questions that parents often ask: When is the right time for a child to own a smartphone? How much screen time is appropriate? How can children be protected from online abuse or cyberbullying? Building on the BEE SECURE publication Bildschirme in der Familie, this two-hour session promotes balanced, age-appropriate approaches to online use and encourages parents to discuss both opportunities and risks with their children.

“Kindersicherung im Internet: praktische Tools, Techniken und Tipps” takes a more technical and solution-oriented approach. Delivered as a webinar, it guides parents through the configuration of security settings and parental-control systems, helping them understand how devices and apps can be adjusted to create safer digital environments for younger users. In the future, this session could further benefit from exploring how AI-powered recommendation systems and automated parental-control tools work, as well as the ethical and privacy issues that arise when algorithms mediate children’s online experiences.

The discussion-based format “Expertentrio: Experten beantworten Ihre Fragen” offers parents the opportunity to engage directly with specialists from BEE SECURE, the child helpline KJT and the crime prevention unit of the Luxembourg Police. By addressing questions about privacy, digital well-being, social media behaviour and online threats, it provides a valuable bridge between professional expertise and family practice. Going forward, this format could also include examples of AI-driven risks, such as deepfake manipulation, synthetic grooming or fake profiles, and offer parents concrete strategies for recognising and reporting such incidents.

Together, these parent-focused initiatives form an essential component of Luxembourg’s broader digital education framework. By integrating AI-related issues more systematically, BEE SECURE can empower parents to guide their children with confidence, foster critical conversations at home and ensure that families remain resilient in the face of an increasingly algorithmic online world.

To maximise inclusivity and accessibility, BEE SECURE should consider offering these parental trainings in additional languages beyond Luxembourgish, German and French. Many families with a migration background may not be fluent in the country’s official languages, yet face the same digital challenges. Providing translated sessions or subtitled online versions in widely spoken languages such as Portuguese or English would ensure that all parents — regardless of linguistic background — can benefit from the guidance and actively participate in promoting safe and responsible internet use at home.

In addition, developing asynchronous online webinars could make the programme even more accessible. Such webinars would allow parents to participate at a time that suits their personal schedules and could be automatically translated into multiple languages using AI-based tools. This approach would significantly expand the reach of BEE SECURE’s parental support, ensuring that high-quality guidance on online safety and AI-related risks is available to all families, irrespective of time constraints or language barriers.

5.2 Ethical considerations for education organisations

5.2.1 Main issues

Oversight & accountability: When oversight processes, such as how educational content is vetted, reviewed, or approved, are not publicly documented, accountability to external stakeholders such as teachers, parents and civil society becomes limited. Without insight into who makes these decisions and on what basis, there is a risk that bias, selective framing, or external influence could go unnoticed. From an ethical standpoint, educational institutions have a responsibility not only to be impartial but also to be seen as impartial. This distinction underscores the importance of procedural transparency: openness about review criteria, expert involvement and revision processes demonstrates integrity and fairness. By making these procedures visible and accessible, institutions reinforce their credibility and ensure that learners and the public can trust that educational resources are developed in an impartial and accountable manner.

Neutrality & advocacy: Ethical questioning lies in recognising that digital literacy education is never entirely free of value judgments, and that discussions on truth, bias, and responsibility inevitably reflect societal norms and priorities. Therefore, the goal is not to present content as neutral “in absolute terms” but to acknowledge the principles guiding educational choices and encouraging critical reflection rather than acceptance.

Empowerment & protection: One problem may also involve the balance between safety (or protection) and autonomy (or empowerment). Parents and schools often use monitoring tools and restrictions to protect youth from online risks, but these measures can also lead to mistrust and limit students’ freedom to explore and learn. Constant surveillance can make young people feel “spied on”, discourage them from seeking help when needed and restrict opportunities for self-expression. Children and young people are entitled to participate fully in the digital sphere, but they are also in need of protection from various forms of harm. Overprotective approaches can restrict opportunities to build digital competence, while overly permissive strategies may increase exposure to social, psychological, or economic risks; different tensions may rise as well between global and local levels of responsibility.

Education & exposure: Educating young people about certain risks carries the inherent danger that they will become aware of potentially harmful AI applications while learning about them. This dilemma is often framed as the “awareness paradox”: Wu et al. (2025) show that increasing AI literacy improves self-efficacy and risk recognition, but note that exposure to examples of harmful uses can trigger curiosity and experimentation if not framed ethically. Similarly, Dubois (2024) applies paradox theory to generative AI in education, emphasising the need to balance transparency with safeguarding to avoid reinforcing discriminatory or harmful practices.

5.2.2 Standards for decisions and trade-offs

Building upon the ethical dilemmas identified above, the establishment of clear standards for decision-making is essential to ensure that digital literacy initiatives like BEE SECURE operate with integrity and fairness. Ethical reflection must evolve from identifying potential risks into defining concrete principles that guide action. Recognising that education in digital ethics and safety is inherently value-laden, these standards must balance competing priorities through deliberate, context-sensitive choices.

First, oversight and accountability demand full procedural transparency as a foundational principle. The ethical concern highlighted earlier — limited visibility into who makes decisions and on what grounds — can be addressed by establishing open documentation and participatory review systems. Publicly accessible information on how educational materials are developed, vetted and approved ensures that decision-making processes remain subject to external scrutiny. Clear disclosure of review criteria, expert involvement and revision procedures reinforces institutional credibility and safeguards against potential bias or external influence. As Floridi et al. (2020) argue, socially responsible digital initiatives must make their objectives and influences explicit to sustain public trust. For BEE SECURE, this includes publishing details about partnerships, funding sources and quality assurance processes, thereby turning transparency into both an ethical safeguard and a practical mechanism for accountability (see MediaSmarts).

Second, as discussed in the first section, digital literacy education inevitably engages with normative issues such as bias, truth, and responsibility. The standard, therefore, should be an explicit acknowledgment of the values that inform educational choices, alongside a commitment to fostering critical reflection among learners.

Third, the issue outlined earlier, between safeguarding young users and respecting their autonomy, requires a proportional and context-sensitive approach. Protective measures such as content filters or parental monitoring are ethically justified only when they serve the learner’s best interest without undermining trust or independence. Excessive surveillance risks cultivating compliance rather than competence, while neglecting protection can expose young people to online harm. To navigate this trade-off, BEE SECURE should adopt the principle of receiver-contextualised intervention (Floridi et al., 2020), tailoring its educational strategies to the capacities, maturity, and needs of learners. Programmes should aim to build digital resilience, the ability to assess and respond to risks independently, rather than relying on restrictive control.

With regard to the awareness paradox, the same principles apply. Child safety frameworks, including the American Psychological Association (APA) Health Advisory (2025), recommend focusing on capacities and principles — such as privacy, consent, and critical thinking — rather than technical “how-to” details of exploitative behaviours. Microsoft’s Safer Internet Day initiatives emphasise interactive learning environments (e.g., Minecraft CyberSafe AI) that teach responsible AI use without glamorising abuse scenarios (Microsoft Education Team, 2025). Together, these perspectives suggest that effective AI safety education should prioritise ethical reasoning, resilience-building, and scenario-based discussions over explicit instructions, thereby mitigating the risk of inadvertently enabling harmful behaviours.

5.2.3 Good practice examples

Several international initiatives offer examples of how ethical dilemmas in digital education and youth protection can be addressed:

  1. MediaSmarts in Canada has established practices to ensure neutrality in educational resources, with oversight by an independent ethics body: https://mediasmarts.ca/about-us/integrity-mediasmarts.
    Examples:
    • “None of our funders exert influence on any of our work, including the content of research and resources”.
    • “MediaSmarts maintains editorial independence over all our owned and co-owned content. […] MediaSmarts does not endorse any products or services and any content that offers guidance on specific digital tools and platforms does not constitute an endorsement”.
    • “We uphold public accountability practices with our funders which include clear guidelines on conflict of interest and sponsorship. We also have a commitment to a community-based research approach which includes a review and approval of our research studies by the Community Research Ethics Office”.
    • “Finally, we value intellectual humility. As a team, we stay open to new evidence and perspectives on identified best practices […] We learn from our network of colleagues and adjust our positions based on the best available evidence”.
  2. klicksafe is a European Union initiative that promotes digital literacy and supports people in using the internet safely, critically and responsibly. It provides independent information, educational materials and training for teachers, parents and young people, while coordinating Germany’s Safer Internet Centre under the EU’s Digital Europe Programme: https://www.klicksafe.eu/en.
  3. The UK Safer Internet Centre (UKSIC) is a partnership of Childnet, Internet Watch Foundation and SWGfL that works to make the internet a safer place through education, support, and reporting services. It coordinates Safer Internet Day in the UK, runs helplines and a hotline for online harms and promotes youth participation through its Youth Advisory Board: https://saferinternet.org.uk/about.

5.3 Recommendations

Based on the analysis of risks, opportunities and international practice, we propose the strategic recommendations below for BEE SECURE’s future work in an AI-driven online environment. To respond effectively to the rapidly evolving digital landscape, BEE SECURE should strengthen its work on AI-related risks while continuing to address the full range of established online risks that remain highly relevant for children, young people and adults.

In light of the rapid proliferation of AI-related trends, BEE SECURE should not develop dedicated trainings or awareness activities for every new technological development, but instead prioritise those issues that are demonstrably relevant for child protection, education and prevention.

For BEE SECURE, the findings on vulnerable groups underscore the need for differentiated and inclusive AI-literacy strategies that reflect Luxembourg’s linguistic and social diversity.

Future activities should focus on multilingual awareness campaigns, early-age interventions, gender-sensitive prevention programmes and professional training on AI-mediated harms such as deepfakes, synthetic sextortion and algorithmic discrimination. Embedding these priorities across all BEE SECURE training formats will help ensure that prevention, counselling and awareness efforts remain responsive to emerging AI challenges while promoting digital empowerment for all children and young people. These priorities are reinforced by the internal staff survey, which highlights similar risk patterns in frontline practice — particularly the heightened vulnerability of girls and young women and the need for multilingual, low-threshold support for families with limited digital or linguistic resources.

Concretely, we propose six interconnected areas of action:

  1. establish comprehensive educational initiatives,
  2. provide guidance for parental and educator involvement,
  3. ensure familiarity with reporting procedures and rights,
  4. create mechanisms for youth participation in policy development,
  5. include ethical reflection in all working areas, and
  6. foster scientific collaboration.

Each area translates the analytical and ethical considerations of the previous chapters into concrete, workable steps for BEE SECURE. While some measures can be implemented in the short term, others are conceived as medium- to long-term developments. All proposed measures build on BEE SECURE’s existing mandate and are designed to complement, not duplicate, the responsibilities of schools, ministries and specialised services.

5.3.1 Comprehensive educational initiatives

In line with international programmes such as Intel’s AI for Youth, and with the AI-literacy and teacher professionalisation needs identified in Chapters 3 and 4, BEE SECURE should extend its educational work from “AI as special topic” towards a cross-cutting competence:

A) Systematically integrate AI into existing BEE SECURE workshops and trainings

Building on the typology of risks (Chapter 2) and the analysis of BEE SECURE’s portfolio (5.1), AI-related content should become a recurring thread in all relevant formats:

  • In primary school formats, include simple “real vs. synthetic” exercises (e.g. distinguishing AI-generated from real photos or voices) to address early-age vulnerabilities.
  • In offers for adolescents, use concrete examples of AI-generated images, videos and texts to discuss manipulation, body image, peer dynamics and opinion formation.
  • In discussion formats with older youth, systematically introduce cases of AI-assisted grooming, sextortion, fraud and algorithmic discrimination.

This does not require entirely new programmes; we need only add a structured “AI layer” within existing formats.

B) Develop an age-graded AI-literacy framework for BEE SECURE

To connect with the KI-Kompass and international AI-literacy frameworks, BEE SECURE should articulate a simple, age-graded AI-literacy framework that can guide all materials and trainings. It should define:

  • core learning goals (e.g. “understand that AI can hallucinate”, “recognise synthetic images”, “know when to distrust a chatbot”);
  • age-appropriate didactic approaches (from playful exploration in primary school to reflective, ethics-based discussion in secondary school);
  • links between AI-literacy and existing competence areas (critical thinking, media literacy, digital civic participation).

This framework can serve as an internal reference point and as a communication tool with schools and partners.

C) Develop new thematic dossiers on cross-cutting AI issues

In addition to updating existing materials, BEE SECURE can create thematic dossiers that bundle central AI-related challenges identified in this White Paper. Priority examples include:

  • “AI and relationships”: – addressing chatbots and virtual companions, parasocial “friendships”, emotional dependency and the blurring of boundaries between human and synthetic interaction.
  • “AI and mental health”: – examining the ambivalent roles of AI-based self-help tools and chatbots, including risks of normalising self-harm content, reinforcing body-image problems or amplifying loneliness.
  • “AI and school”: – focusing on homework, grading and exams, the changing role of teachers in AI-mediated learning, and fair, transparent rules for student use of generative systems.

These dossiers should synthesise research findings from this report with concrete pedagogical implications and be complemented by youth-friendly formats (short explainers, videos, comics) that can be used directly in schools and youth work.

D) Establish a recurring “AI & Youth” campaign strand

Building on BEE SECURE’s campaign experience and the international good practice of using gamification and simulations (5.3.1), BEE SECURE should introduce a recurring campaign strand explicitly devoted to AI. Over a defined period (e.g. annually or every two years), it would focus on:

  • deepfakes and synthetic sexualised content,
  • automated scams and voice-cloning fraud,
  • AI-assisted hate, harassment and disinformation.

The campaign should be co-created with young people (see Section 5.3.4) and linked to Safer Internet Day and other national events, ensuring that AI becomes part of mainstream online-safety discourse.

Where possible, new or revised AI-related formats should be accompanied by proportionate evaluation mechanisms (e.g. short pre/post feedback instruments) to assess reach, relevance and perceived impact.

E) Use regular foresight workshops to keep offers up to date

To anticipate rather than merely react to new AI risks, BEE SECURE should hold regular foresight workshops (e.g. every six months) with researchers, youth workers, law enforcement and other stakeholders. These workshops would:

  • map emerging AI-related phenomena (e.g. new deepfake apps, AI in games, political micro-targeting);
  • assess their relevance for children and young people in Luxembourg;
  • feed directly into the revision of workshops, materials and campaigns.

5.3.2 Provide guidance for parental and educator involvement

Echoing the logic of Intel’s AI for Citizens programme and the findings on teacher capacity and family support (Chapters 3 and 4), BEE SECURE should equip adults who mediate children’s digital lives with concrete tools and guidance:

A) Expand trainings for teachers, educators and parents to include AI didactics

Existing BEE SECURE trainings can be complemented with modules that focus on how to integrate AI into education and parenting in a reflective way:

  • designing tasks where students critique and revise chatbot outputs, instead of simply copying them;
  • developing fair and transparent criteria for assessing AI-assisted homework and projects;
  • discussing proportionate responses to AI-supported cheating and plagiarism that prioritise learning and responsibility over punishment.

This directly implements the ethical standards of Chapter 5.2 by focusing on judgement, proportionality and transparency.

B) Strengthen gender-sensitive and diversity-sensitive approaches in existing formats

On the basis of Chapter 4.3 and Table 2 on vulnerable groups, BEE SECURE should sharpen its focus on specific risk constellations within current offers for adults:

  • include modules on AI-mediated sexualised violence (deepfakes, “nudifying” apps, synthetic sextortion) in trainings for professionals and parents, especially those working with girls and young women;
  • develop complementary components for boys and young men on consent, bystander roles and responsibility when creating or forwarding synthetic sexual content;
  • provide guidance for professionals working with young people with disabilities or mental-health vulnerabilities, for whom AI tools may create specific risks (e.g. stigmatising content, self-harm forums mediated by AI).

C) Provide multilingual guidance and checklists for families

To reflect Luxembourg’s linguistic diversity and the survey’s finding on families with limited resources, BEE SECURE should develop concise, multilingual guidance on “safe AI use in the family”:

  • short checklists for parents on talking about AI, setting boundaries for image sharing, and responding to deepfakes;
  • easily accessible factsheets outlining children’s and parents’ rights in relation to AI-generated content and data protection;
  • modular webinars addressing key AI-related risks and protective strategies.

To ensure low-threshold access, these materials should be made available in flexible, asynchronous formats, particularly for families that are difficult to reach through traditional formats. Where appropriate, AI-based translation and subtitling tools can be used to provide additional language versions.

These materials can be promoted through schools, community organisations and health services.

D) Pilot a curated “digital advisor” (AI chatbot) as a guidance tool

BEE SECURE can pilot a carefully curated, privacy-preserving AI-based “digital advisor” on its website or app. For parents and educators, such a tool could:

  • answer frequently asked questions about AI and online risks in accessible language;
  • point to relevant BEE SECURE resources, trainings and campaigns;
  • clearly signal situations where professional human support (Helpline, KJT, psychological services) is necessary.

As a ministerial licence for platforms such as Fobizz is already in place, such a chatbot could be created and tested in a low-threshold manner using existing tools, without requiring complex technical development. This would allow BEE SECURE to focus on content, guidance and safeguards rather than on building new technical infrastructure.

The chatbot should operate within clear ethical and quality safeguards, be transparently labelled as informational only, and undergo regular expert and youth review. The digital advisor is intended exclusively as an initial orientation and signposting tool and does not replace personal counselling or professional judgement.

5.3.3 Ensure familiarity with reporting procedures and rights

To make protection effective, children, young people, parents and professionals must understand how to report harmful content and what rights they have. At the same time, BEE SECURE’s own procedures must be adapted to AI-mediated harms and make careful use of AI for risk detection (as highlighted in 5.3.1):

A) Adapt Helpline and Stopline procedures to AI-related cases

BEE SECURE should refine its internal taxonomies and workflows so that AI-related harms are clearly visible in practice:

  • introduce specific categories such as deepfake harassment, voice-cloning fraud, chatbot-mediated grooming and synthetic CSAM;
  • provide targeted training and supervision for counsellors on these phenomena, including psychological impacts and collaboration with law enforcement.

This will allow the Helpline and Stopline to address new forms of harm while preserving their core relational and trust-based approach.

B) Make reporting channels and rights visible in all AI-related materials

All materials that address AI risks should, wherever appropriate, include clear information on:

  • how to contact BEE SECURE Helpline and Stopline;
  • how to report unlawful content to platforms and authorities;
  • which rights children, young people and parents have (e.g. removal of unlawful content, protection against discrimination, data-protection rights).

This ensures that AI-related education is systematically linked to concrete avenues of support and redress.

C) Establish an AI-specific monitoring and early-warning strand

BEE SECURE’s Radar and internal monitoring processes should be expanded with an AI-specific strand:

  • integrate question blocks on AI use, exposures and perceived risks into surveys;
  • systematically analyse AI-related Helpline and Stopline cases;
  • compile an annual “AI incidents and perceptions” section in public reporting.

Subject to strict data-protection and ethics safeguards, BEE SECURE may also explore AI-assisted analysis of aggregated and anonymised data to identify emerging patterns (e.g. new scam types, new deepfake practices), thereby turning frontline experience into an early-warning system.

D) Adopt an ethics and quality framework for AI-related work

Drawing on the MediaSmarts example and the normative standards developed in Chapter 5.2, BEE SECURE should formalise a short ethics and quality framework for its AI-related activities, covering:

  • editorial independence from funders and commercial partners in all content and tools;
  • transparent disclosure of partnerships, sponsorships and potential conflicts of interest;
  • a clear statement that any cooperation with platforms does not imply endorsement of their products;
  • commitments to public accountability and to revising positions in light of new evidence and youth feedback.

This framework should apply to all AI-related initiatives — from factsheets and trainings to monitoring and any AI-based tools — and can be published on BEE SECURE’s website. This framework does not introduce new regulatory standards but operationalises existing ethical principles for BEE SECURE’s educational and preventive work.

E) Develop and rehearse crisis plans for AI-related “worst-case” scenarios

Findings from the internal staff survey indicate that, despite overall confidence in handling AI-related inquiries, advisors encounter uncertainties in complex or ambiguous situations. These insights underscore the need for clearer crisis plans, decision trees, and standard operating procedures (SOPs) to guide responses in potential worst-case scenarios involving AI-generated harm or manipulation. BEE SECURE should work with schools, police and other stakeholders to create specific standard operating procedures for AI-related crises, such as:

  • viral circulation of a deepfake sexual video of a student;
  • synthetic CSAM involving Luxembourg contexts;
  • AI-mediated self-harm or suicide “advice”.

These SOPs should define escalation paths, communication strategies and support measures for affected children, families and schools. Regular tabletop exercises will help test and refine them.

These efforts should be aligned with relevant European frameworks such as the Digital Services Act and the emerging AI regulatory landscape.

5.3.4 Create mechanisms for youth participation in policy development

As argued throughout this report, young people are not only at risk but also key experts in AI-mediated digital cultures. In line with international good practice (e.g. the UK Safer Internet Centre Youth Advisory Board), BEE SECURE should institutionalise youth participation:

A) Establish a permanent “AI Youth Advisory Group”

BEE SECURE should set up a youth advisory group focused on AI that:

  • reviews new AI-related materials, campaigns and digital tools (including any chatbot);
  • tests their clarity, relevance and tone;
  • regularly reports back to BEE SECURE staff and, where appropriate, to policymakers.

This group can be linked to existing youth councils to avoid duplication and ensure democratic legitimacy.

B) Build an “AI Safety Ambassadors” peer-education programme

Following peer-to-peer models highlighted in 5.3.1, BEE SECURE can train selected students and young people as AI Safety Ambassadors, who then:

  • act as peer multipliers in schools, youth centres and events;
  • co-design and co-facilitate workshops and awareness activities;
  • serve as approachable contact points for peers seeking guidance on AI-related issues.

Participation can be recognised through certificates or links to civic engagement programmes, increasing motivation and sustainability.

C) Co-create campaigns and educational content with young people

Youth participation should extend beyond consultation to co-creation, particularly in the “AI & Youth” campaign strand:

  • involve young people in designing slogans, visuals, videos and social media content;
  • pilot new workshop formats in schools and youth centres and revise them based on youth feedback;
  • systematically incorporate feedback from youth structures into updates of materials and campaigns.

D) Involve youth structures in AI policy dialogue and evaluation

Finally, BEE SECURE should collaborate with youth councils, school student bodies and other youth organisations to:

  • include youth perspectives in national and European debates on AI governance and youth rights;
  • invite young people to evaluation workshops on BEE SECURE’s AI-related initiatives;
  • ensure that their recommendations are taken into account in strategic decisions.

5.3.5 Include ethical principles in all working processes

A) Ethical reflection

AI-related educational initiatives should incorporate structured moments of ethical reflection throughout their lifecycle. Rather than treating ethics as a one-time compliance step, institutions should periodically reassess whether tools and practices continue to align with educational goals, developmental needs, and emerging evidence.

B) Ethical AI use

Ethical AI use should be framed as a shared responsibility between institutions, educators, parents, and learners. Educational initiatives should clarify the respective roles of each actor, avoiding the implicit transfer of responsibility to adolescents themselves. Institutions should establish mechanisms that allow adolescents to reflect on and provide feedback about their experiences with AI-mediated tools and content. This feedback should inform revisions and adaptations, recognising young people as stakeholders rather than passive recipients. Participatory input strengthens legitimacy and helps identify impacts that may not be visible to adults.

C) Future Orientation

Institutions should prepare adolescents not only for current AI applications but also for plausible future developments. This includes fostering anticipatory awareness and helping young people recognise how ethical risks may evolve, rather than focusing solely on present tools.

5.3.6 Strengthen evidence-based prevention through scientific collaboration

Scientific collaboration can serve as a strategic enabler for BEE SECURE’s long-term, evidence-based development. Rather than constituting a separate programme, cooperation with academic partners allows existing data, expertise and prevention activities to be systematically analysed, evaluated and translated into practice-relevant insights. In this sense, scientific collaboration provides a foundation for informed policy decisions, continuous improvement of prevention strategies and international visibility. One promising option is a long-term cooperation with the University of Luxembourg. BEE SECURE continuously generates data on digital risks (helpline/stopline statistics, radar surveys, training feedback). This data is highly relevant for empirical research on media education, youth online behaviour and AI risks. On the other hand, the University of Luxembourg has expertise in digital ethics, educational technology and teacher education. Joint projects can develop and evaluate evidence-based prevention strategies. The university also provides a platform for systematically reaching target groups such as prospective teachers, social pedagogues and educators — e.g. through integration into study modules or MOOCs. Possible areas of cooperation include:

  • Data-sharing framework: Development of a legally compliant cooperation model for anonymised data (Radar, Helpline, Stopline), including GDPR-compliant governance.
  • Joint research projects: Joint third-party funding applications (e.g. FNR, EU Horizon) on topics such as ‘AI Literacy in Schools,’ ‘Deepfake Detection Skills’ or ‘Impact of AI Companions on Youth Well-being.’
  • MOOCs & Teacher Training: Development of open online courses on ‘AI & Digital Security’ for teacher training students and continuing education staff, based on BEE SECURE content and university didactic expertise.
  • Evidence-to-Practice Lab: Establishment of an ‘AI Safety Education Lab’ as an interface between research, practice and policy – piloting innovative teaching formats and evaluating their effectiveness.
  • Youth Advisory & Participatory Design: Involvement of school pupils and students in co-design processes for new awareness campaigns and tools.

Annotate

Next Chapter
References
PreviousNext
CC BY-SA 4.0
Powered by Manifold Scholarship. Learn more at
Opens in new tab or windowmanifoldapp.org