Hi PH — Alena here, the one person behind this. Article 50 of the EU AI Act has applied since August 2: visitors must be told when they're talking to a machine. Everything answering "are we OK?" was either a questionnaire that takes your word for it or a widget grading its own homework. Nothing just looked, the way a visitor looks. So this does: real browser, dismisses the consent wall, opens the chat, reads the first message. On your own verified site it goes further — it asks your assistant whether it's an AI and reads the answer. Findings are detected / not detected / could not verify. It will never call you "compliant" — that conclusion belongs to your lawyer; what you get is a dated, hashed record your lawyer can actually use. Earlier this month I published a study of 1,088 detector-flagged, EU-facing sites with the scanner — including its own measured error rates, because a compliance tool that won't audit itself shouldn't audit you: https://disclosureproof.com/rese... The diagnosis is free, no signup. Paid is the sealed Evidence Pack (€79) and monitoring that emails you when your disclosure silently drifts. Ask me anything — replies reach me, not a queue.

DisclosureProof

See your site the way an AI-law inspector would

See your site the way an AI-law inspector would