Loading page
About this tool
Practise spotting synthetic media and misleading AI claims, with scored questions on provenance, detectors and verification.
The AI Media Literacy Quiz scores how well you can judge whether an image, video, voice clip or claim was produced or distorted by AI, using scenario questions drawn from how verification actually works — C2PA content credentials, reverse image search, out-of-band confirmation and primary-source checking. Every answer is explained, so a wrong choice teaches the underlying rule rather than just deducting a point. It is built for teachers, journalists, moderators and anyone briefing a team on synthetic media.
Open AI Media Literacy Quiz on AltFTool — it loads instantly in your browser.
Choose "Questions in this round" — 5, 8 or 12 — and set the "Question set" number; the same set number always returns the same questions in the same order, so a class can take an identical round.
Answer each "Question N of N" by selecting one of its options, then press "Check answers"; Shuffle draws a fresh set and Reset clears your answers, both behind a confirmation.
"Your score" shows the percentage with Answered, Correct, Wrong, the 70% pass mark, Pass or Below pass mark and your Weakest topic, and every question reveals its explanation; "Copy score" copies the summary.
Each question ends with the reasoning and the real-world failure mode behind it.
A set number fixes the question order, so a class or team can all take the identical round.
Scores split across provenance, detector limits, verification workflow, AI claims and fraud so you know what to study.
No. Detector outputs are probability estimates that shift with compression, cropping and screenshotting, and text detectors in particular have documented false positives that fall hardest on non-native English writers — OpenAI withdrew its own text classifier for low accuracy. Treat any detector score as one weak signal that needs corroboration, never as proof.
C2PA attaches a cryptographically signed manifest recording the capture device, generator and edit history of a file. When present and valid it is strong evidence, but absence proves nothing: the metadata is easily stripped and most social platforms remove it during upload.
Start with a reverse image search to see whether the picture appeared earlier in another context — recycled real photos with false captions are far more common than deepfakes. Then look for the original file from a named photographer and independent images of the same scene, which together beat any visual artefact check.
Verify out of band. Hang up and call back on a number you already hold, agree a family code word in advance, and require dual authorisation plus a separate channel for payment requests at work — in a widely reported 2024 Hong Kong case a finance employee transferred roughly US$25 million after a video call in which all the other participants were synthetic.