Seeing Is No Longer Believing

Executive Summary

Media and information literacy is a key tool for recognising and countering false information online. However, generative AI presents new challenges: manipulated photographs, synthetic voices, videos and written material can all appear convincing. This raises an urgent question for policymakers and literacy professionals: what kind of literacy is needed when seeing is no longer believing?

This brief focuses on young Lebanese students who encounter AI tools and online information in their studies and everyday lives. It draws on a small exploratory survey of 36 students and four in-depth interviews. All 36 survey respondents said they had encountered content in the previous three months that they suspected was AI-generated, manipulated or false. That is not the same as confirming that the material was AI-generated misinformation. Among the survey respondents, 64% selected ignoring suspicious content as a usual response, while 17% selected reporting it.

The issue is bigger than a lack of awareness. Some students described ways to check suspicious posts, but interviews also revealed uncertainty about what to trust and whether reporting would make a difference. Platform design matters, too: recommendations, engagement signals and reporting systems can shape what people see and how credible it seems. These findings offer early insights, not a representative picture of Lebanese students.

The brief recommends four connected areas of action:

• Practical AI literacy. Schools, universities and civil society organisations should use real examples to teach verification and responsible sharing.
• Platform literacy. Students should understand how algorithms, recommendations, likes, comments, views and reporting systems influence the content they encounter.
• Clearer platform safeguards. Platforms should make reporting straightforward, improve feedback, and use human review when language or local context matters.
• Shared responsibility. Lebanese policymakers, educators, journalists, fact-checkers and technology platforms should develop a coordinated approach that protects access to information and freedom of expression.

Students should not carry the whole burden. A safer information environment requires prepared users, accountable platforms, responsible institutions and clear public guidance.

Why This Matters Now

Governments, educators and civil society have long used media and information literacy to help people resist misinformation. It is increasingly linked to trust, democratic participation and information resilience.¹ Generative AI adds a new question: when young people use AI and social media as part of daily life, are they prepared to judge suspicious material and decide what to do about it?

This brief focuses on Lebanese students and their encounters with social media and generative AI. Tools such as ChatGPT, Gemini, Claude and NotebookLM can be used for studying, writing and research. Familiarity with such tools, however, does not necessarily prepare users to recognise or respond to misleading content.

The challenge is not only whether young people can spot a fake image or video. It is also whether they know how to verify suspicious content, when reporting may be useful, how to avoid spreading it and how platforms shape what appears in their feeds.

The core concept is AI literacy. Researchers define it in different ways: some emphasise technical understanding, while others focus on responsible use, ethics and critical thinking.^{2, 3} This brief takes a practical approach. AI literacy means having the knowledge, skills and judgement to question AI-generated material, assess its reliability, understand potential risks and act responsibly. Platform literacy supports that approach by examining the algorithms, recommendations, likes, comments, shares, reporting systems and moderation rules through which content circulates.⁴

The purpose is to understand how young Lebanese students identify, evaluate and respond to suspected AI-generated or manipulated material on social media, and what their experiences may suggest for educators, platforms and policymakers. Students may be aware of AI without feeling prepared for everything it changes about the information they see, trust and share.

Approach and Results

This policy brief is informed by exploratory research carried out with Lebanese students in summer 2026. A survey was circulated through online student groups using snowball sampling, meaning that respondents could share it within their networks. It produced 36 responses and was complemented by four in-depth interviews.

The survey asked about students’ use of AI tools, trust in online information, familiarity with AI, encounters with content they suspected was AI-generated, manipulated or false, ways of recognising and verifying such material, reporting habits, understanding of platforms and views on responsibility. The interviews explored experiences of encountering suspicious content and what students did next.

Because the sample was small and recruited through student networks, it cannot represent all students in Lebanon. The findings describe self-reported experiences and actions; they do not test participants’ ability to detect misinformation or independently establish that the content they saw was false or AI-generated.

Repeated encounters with suspicious content

All 36 respondents said that, in the previous three months, they had seen online content they suspected was AI-generated, manipulated or false. Thirty-one (86%) answered “many times”; five (14%) answered “a few times.” For this group, suspected manipulation was a familiar part of the online environment.

What students do when they see it

Many respondents reported taking steps to check suspicious content. Nineteen (53%) said they search for another source, 17 (47%) read comments and 14 (39%) check the account or page. Yet 23 (64%) also selected “ignore it”, the most frequently chosen option. Only six (17%) selected “report it”. Respondents could choose more than one action, so these behaviours are not mutually exclusive. Reading comments may offer useful context, but it is not, by itself, a reliable verification method.

The interviews help explain this pattern. Participants described being more likely to check material that felt serious—politics, war, major news, university or work—than entertainment such as memes or celebrity posts. Social media was also described as a place to relax or pass time, rather than somewhere users feel obliged to investigate every post.

“When people scroll, they are using minimal energy. They are not there to police the platform.”

Visual clues were another cue. Interviewees mentioned strange body movements, unnatural voices, lip-syncing mistakes, blurry backgrounds, fake-looking images or content that appeared “too perfect”. They also described reading comments, checking reposts, asking friends and using search engines. These habits may help, but realistic synthetic content makes visual judgement alone increasingly uncertain.

“Sometimes I feel helpless. I cannot know if it is fake or not.”

These experiences echo a wider concern in research on AI use and literacy: familiarity with tools does not automatically mean that people feel confident evaluating online information or deciding how to respond.⁵ But these interviews should not be treated as proof of a general skills deficit. Choosing not to report or verify every post may also reflect doubts about effectiveness, the time involved or the importance of the content.

Media literacy and information resilience are prominent in policy discussions in the Nordic countries and Europe.⁶ European digital competence frameworks now integrate AI more explicitly, while debates about transparency and responsibility continue.⁷ The Lebanese experiences described here provide a small, grounded starting point for considering what practical support might look like locally.

Recommendations: From Scrolling to Responsibility

Students may know that AI exists and encounter its outputs often, but that does not mean they always have the tools or support to deal with misleading, emotional, political or harmful content. AI literacy is necessary, but it is not sufficient. Students cannot be expected to verify and report everything they see. The recommendations below aim to help them respond more confidently while asking platforms and institutions to do their part.

1.Teach practical AI misinformation literacy

Schools, universities and civil society organisations should add real examples to AI literacy education, not rely on general awareness alone. Training should be short and practical. Students should learn how to pause before sharing, compare independent sources, check an original account, consult professional fact-checkers and recognise when content cannot be verified. Examples should include AI-generated images, synthetic voices and deepfake videos, as well as ordinary misleading posts.

2.Add platform literacy to AI education

Platforms do not simply host information; their systems help decide which content reaches users. Explaining how recommendations, algorithms, likes, comments, views and reporting mechanisms affect what students see and trust should be a priority. Before believing or sharing a post, students should be encouraged to ask: Why am I seeing this? Who posted it? Is popularity the same as credibility? Can comments help—and can they also mislead?

3.Make reporting easier and more trusted

Reporting AI-generated or manipulated content should be straightforward. Platforms should offer clear options for reporting misleading synthetic content and provide understandable feedback about what happens next. Some interviewees expressed little confidence that reporting would lead to action:

“I don’t report because I don’t trust Meta.”

Reporting tools should fit how young people actually use social media. A student scrolling casually is unlikely to complete a long or confusing process. Reporting should be fast, clear and credible, while allowing users to distinguish harmful deception from benign or clearly labelled AI-generated content.

4.Strengthen platform detection, labelling and expert review

Users should not be the only ones expected to flag misleading synthetic content. Platforms should improve the detection, labelling, review and reduction of harmful AI-generated misinformation. This does not mean removing all AI-generated content. The focus should be on misleading material that risks causing harm, spreading false claims or damaging reputations, with proportionate safeguards for legitimate expression.

Automated systems alone cannot always understand context. Research on human and AI-assisted moderation illustrates why human judgement remains important in digital spaces.⁸ In Lebanon, review can benefit from Arabic-language expertise and knowledge of local politics and crises, potentially through cooperation with Lebanese fact-checkers, journalists, digital rights groups and media literacy organisations.

The aim is not censorship. It is to make harmful or deceptive content easier to identify, assess and address, while ensuring that reporting and moderation decisions are transparent and open to challenge.

5.Treat AI misinformation as a shared responsibility

The students interviewed did not see the problem as one for individuals alone. They called for AI education, ethics courses, platform filters, clearer reporting, stronger rules and greater pressure on technology companies. These ideas connect with broader educational and policy efforts, including European work on digital competence and AI literacy.^{7, 9}

Lebanon can draw on these approaches while designing measures for its own circumstances. Universities, media organisations, fact-checkers, platforms and policymakers should work together on accessible resources, Arabic-language and Lebanon-specific examples, and guidance for high-risk content during conflict, crises and political tension. Given the limited leverage of any one country over global platforms, coordination across institutions and with international partners may be more effective than isolated requests.

The final message is simple. The students in this exploratory study were familiar with the possibility of AI-generated misinformation, but some were unsure how to act, when to report or whom to trust. Literacy should help users respond more confidently; it should not make them responsible for fixing the entire system. A safer information environment needs prepared users, accountable platforms, responsible institutions and public rules that protect trust without limiting expression.

References

1. [1]Hamzeh, D. (2023). The Role of Media Literacy in Countering Misinformation. Thesis, Lebanese American University. https://doi.org/10.26756/th.2023.634
2. [2]Hillman, V., Holmes, W., & Duarte, T. (2025). A Rapid Review of AI Literacy Frameworks, with Policy Recommendations. Report prepared for the Royal Society. https://royalsociety.org/-/media/policy/projects/ai-in-education/hillman-et-al-a-rapid-review-of-ai-literacy-frameworks.pdf
3. [3]Chiu, T. K. F. (2025). AI literacy and competency: Definitions, frameworks, development and future research directions. Interactive Learning Environments, 33(5), 3225–3229. https://doi.org/10.1080/10494820.2025.2514372
4. [4]Pangrazio, L. (2026). The (im)possibility of AI literacy. Learning, Media and Technology, 51(1), 1–7. https://doi.org/10.1080/17439884.2026.2615553
5. [5]Brown, R., Sillence, E., & Branley-Bell, D. (2025). AcademAI: Investigating AI usage, attitudes, and literacy in higher education and research. Journal of Educational Technology Systems, 54(1), 6–33. https://doi.org/10.1177/00472395251347304
6. [6]Nordic Council of Ministers. Nordic Media Literacy Survey (TemaNord 2025:572; publication released January 2026). https://doi.org/10.6027/temanord2025-572
7. [7]Cosgrove, J., & Cachia, R. (2025). DigComp 3.0: European Digital Competence Framework, fifth edition. Publications Office of the European Union. https://doi.org/10.2760/0001149
8. [8]Waterschoot, C. (2025). Governing the “Third Half of the Internet”: The dynamics of human and AI-assisted content moderation. In J. van Dijck, K. van Es, A. Helmond, & F. van der Vlist (Eds.), Governing the Digital Society: Platforms, Artificial Intelligence, and Public Values (pp. 63–81). Amsterdam University Press. https://doi.org/10.5117/9789048562718_CH03
9. [9]OECD & European Commission. (2026). Empowering Learners for the Age of AI: An AI Literacy Framework for Primary and Secondary Education. OECD Publishing. https://doi.org/10.1787/65cd27d4-en