AI or Authentic?
AI or Authentic?
Check social media images & videos for AI - free, honest, with a shareable certificate.
⚡ Faster checks: iPhone shortcut
Honest assessment - result in about 30-60 s.
No checks on this device yet.
Can you guess today’s case? Build your streak!
AI or Authentic? inspects the media itself - not just a platform label.
No detection is 100% certain. When the evidence isn't enough - e.g. with heavily compressed reposts - we deliberately say "inconclusive" instead of guessing. Better honestly unsure than wrongly accusing someone. More on this under Methodology & limits.
How to spot AI content - all topics at a glance.
The EU's AI labeling rules are in force: 7 answers you need now
AI video generators in 2026: Veo, Kling, Runway - and what is left of Sora
How to Spot AI Images: Telltale Signs of AI-Generated Photos
We do not rely on a single model but combine several complementary signals: checks for provenance records (C2PA / Content Credentials) and watermarks, our own trained model for videos, an ensemble of image forensic detectors, and a calibrated fusion of this evidence into a transparent result. Verified provenance outweighs any pure model score.
There is no hundred percent reliable detection, especially on heavily compressed social-media reposts. That is why we honestly return “inconclusive” when the evidence is not enough, instead of guessing. The result is always an assessment with reasoning, not a blind percentage verdict.
Images and videos, either as a direct upload or as a link (e.g. TikTok, YouTube). Instagram and Facebook links often cannot be loaded because of the login wall; in that case it is best to save the media and upload it directly.
The checked original is deleted after the analysis and not stored permanently. The check only happens after your explicit consent. Details are in the privacy policy.
Yes, checking images and videos is free to use.
When the available evidence points neither clearly to real nor clearly to AI-generated, we honestly report it as “inconclusive” instead of faking certainty. That is a deliberate result, not an error.
