Media Literacy: Teaching Students to Verify What They See

You come out of this week with a verification routine your students can actually run, and a written rule for what happens when it fails.

• Monday — you sort a set of real and generated items before any training, and find out how good you actually are.
• Tuesday — the procedure: provenance, lateral reading, tracing an image back. What still works.
• Wednesday — what EU law now requires generated content to carry, what that marking misses, and why detectors are not the answer.
• Thursday — you turn the procedure into something a class can do in ten minutes.
• Friday — a live clinic on genuinely hard items, then you write your school’s rule for the ones nobody can settle.

This is not a course about spotting bad headlines. It is about what a teacher does when a student brings in something that cannot be resolved by looking harder.

In October 2025 UNESCO published the first global mapping of media and information literacy across all 194 of its member states. The figure that travelled was that 43% of countries have built media and information literacy into formal education. For a European teacher that is the wrong number to look at. In Europe and North America the figure is 91%. Media literacy is almost certainly already in your curriculum.

So the problem is not absence. It is that what is being taught was designed for a world in which producing convincing false material took time, money and skill. That stopped being true, and the law has moved in response.

Since 2 August 2026, Article 50 of the European Union’s Artificial Intelligence Act has required providers of systems that generate synthetic audio, image, video or text to mark that content in a machine-readable format, detectable as artificially generated or manipulated. That is worth teaching. It is also narrower than it first sounds, and the limits matter more than the rule: systems already on the market before that date have until 2 December 2026, and content where AI only assisted with standard editing falls outside the obligation altogether. The obligation sits with the provider of the system, not with the school. A student who knows only the first half of that will trust a mark that was never required to be there.

The week works inside UNESCO’s own curriculum for educators, and says which parts. Module 4 covers competencies to tackle misinformation, disinformation and hate speech; Module 11 covers artificial intelligence, social media and media and information literacy. Those two carry the week, with Module 3 on the information cycle and Module 7 on how media and technology shape content supporting them. The curriculum has fourteen modules. This course works in four, and the first day says which ten it leaves out.

Monday is deliberately uncomfortable. Before any teaching, participants sort a mixed set of real, generated and edited items and record how confident they were about each. The column that matters afterwards is not the wrong answers but the confident wrong answers, and that column sets the week’s priorities.

Tuesday builds the procedure that survives all of this: establishing provenance, reading laterally rather than staring harder, and tracing an image or a claim back to the first place it appeared. None of it depends on being able to see that something is false, which is exactly why it still works when seeing fails.

Wednesday deals with the tool teachers ask about first. Detectors return a score, and a score is evidence rather than proof; a class taught to trust one has been taught to outsource the judgement the course exists to build. What replaces it is the marking requirement, its two gaps, and the habit of asking where something came from rather than whether it looks right.

What travels home is deliberately small: a routine a class can run in ten minutes, and one page setting out what a teacher does with an item nobody in the room can settle — who it goes to, what is said to the student in the meantime, and what gets recorded.