Seeing is not the same as verifying. A visual can be AI-generated, edited, cropped, old, taken somewhere else — or completely authentic but attached to a false caption. You will investigate one visual claim and create a case file showing how context and provenance can be checked. By the end of this challenge you […]
Scan to open this challenge on any device
Old photos, cropped videos and real images with false captions can be as misleading as AI-generated media.
Signs, weather, shadows, landmarks, uniforms, language, vehicle plates and editing cuts can all create searchable clues.
Verification often means going backwards: who posted it earlier, what was the original caption and does the date/location match the new claim?
Use reverse-image search where available, or search distinctive visual details and wording. Find one earlier credible use and compare date, place and caption.
Use multiple methods: reverse-image search, keyframe/still search, landmark or text clues, and independent reporting. Build a short timeline of how the media appeared or changed.
Visual anomalies can be useful clues, but compression, editing and ordinary camera artefacts can also look strange.
Your case file should show the claim, clues, search path, evidence and verdict — not just the answer.
Show the claim, clues, trace-back evidence and verdict on one page.
Walk the audience through claim, checks, original context and verdict.
Screen-record or narrate the verification path without exposing private accounts.
The next challenge zooms out from individual posts to whole feeds: what different people see and how that shapes perception.
Which verification method was most useful in your case?
How could this process help when checking breaking news or viral clips?
How might provenance tools or content credentials change visual verification in the future?
Your Step 5 output is the main evidence. Keep to the stated size or time limit so the focus stays on quality.
Choose Your Tools
Use the tools available to you. The quality of the checking and reasoning matters more than the software.
Use reverse-image search where available, or search distinctive details from the image. Check earliest/contextual uses rather than only visually similar copies.
Look for the original upload, longer clip, date/location clues and independent reporting. Treat AI-detection tools as signals, not proof.
Use a case file, annotated image board, slides or short verification video.
Your Step 5 output is the main evidence. Keep to the stated size or time limit so the focus stays on quality.
Choose Your Tools
Use the tools available to you. The quality of the checking and reasoning matters more than the software.
Use reverse-image search where available, or search distinctive details from the image. Check earliest/contextual uses rather than only visually similar copies.
Look for the original upload, longer clip, date/location clues and independent reporting. Treat AI-detection tools as signals, not proof.
Use a case file, annotated image board, slides or short verification video.
Choose Your Tools
Use the tools available to you. The quality of the checking and reasoning matters more than the software.
Use reverse-image search where available, or search distinctive details from the image. Check earliest/contextual uses rather than only visually similar copies.
Look for the original upload, longer clip, date/location clues and independent reporting. Treat AI-detection tools as signals, not proof.
Use a case file, annotated image board, slides or short verification video.