Article 50 applies now. See what your site shows a visitor.
Since 2 August 2026 you have to tell people when they are talking to an AI, mark AI-generated media so a machine can detect it, and visibly label deepfakes. Point this scanner at any public URL and see, clause by clause, what we can and cannot observe on your live site right now.
No signup, no card, no crawl of the rest of your site. About 20 seconds. We load the page in a real browser and inspect what a visitor actually receives — not just the HTML you serve.
We report what we can observe and say unverified when we cannot observe it. A false accusation is worse than a missed finding. What we refuse to guess at.
What gets checked
AI disclosure on chat & voice
Detects assistants and voice agents, then checks for a disclosure a user can actually see before their first message — in any of the 24 official EU languages.
Art. 50(1), 50(5)Machine-readable provenance
Inspects served images, video and audio for C2PA Content Credentials at the container level, plus XMP and IPTC provenance assertions.
Art. 50(2)Marks stripped in delivery
Finds media that lost its provenance manifest somewhere between generation and the browser — the most common way marking silently fails.
Art. 50(2)Unlabelled deepfake media
Flags synthetic-likeness content served without a persistent, human-visible label.
Art. 50(4)