The Fake of the Week

Seven headlines about AI in schools. Can your family tell which hold up?

Three of these were made up. Three are real — and distorted. One holds up under checking. About six minutes together.

Three headline cards. The first carries a verification tick. The second has a genuine official seal but a date struck through — a real source behind a wrong claim. The third has empty dashed boxes where a name, a date and a case number should be.

The middle one is the hard case: a genuine source, a real seal — and a detail quietly out of date. That is what most people miss.

The thing most people get wrong. Dangerous misinformation is rarely invented from nothing. Far more often it is a real study, a real law or a real report, with the qualifier removed — the date, the subgroup, the "only for some people". That is why this has three answers instead of two.

Pick a level

Choose a level to begin.

Your family scored
0 of 7

The signals worth remembering

  • No name, no date, no number. A real ruling has a case number. A real directive has a document number. If nothing can be looked up, nothing can be checked.
  • "Real source" is not the same as "real claim." The most dangerous item here had four genuine official sources and was still wrong — and wrong in both directions: the duty it meant was deferred to 2027, while other duties in the same law already applied.
  • Watch for the missing qualifier. "Teachers save five hours" was really "5.9 hours, for the 3 in 10 who use it weekly, from memory rather than measurement."
  • One story is not three stories. The same wire report republished by three outlets looks like corroboration and is not.
  • Two of these were the same story. The Massachusetts deepfake item appeared twice — once stretched into a national trend, once told accurately. Same evidence, different claim. That pair is the whole exercise in miniature.

These four claims are circulating right now.

Every headline in this exercise was checked against its sources on 5 September 2026. Two named real laws and real studies, and still got them wrong.

Would your child's school be able to tell the difference? Most cannot yet — not through carelessness, but because nobody has ever mapped it.

See where a school actually stands

Fifteen questions, five minutes, nothing leaves your browser. Built for the person responsible for AI at a school — share it with yours.

Questions to ask your school

Every item here comes from a real verification run, not from examples invented for the exercise. Where a claim is false, the falsity is documented and the evidence is shown. A tool that teaches people to spot fabrication should not fabricate its own material.