The EU AI Act's disclosure rule quietly kicked in most "compliance" tools just slap a banner on it

A few months ago I started digging into Article 50 of the EU AI Act the rule that says users need to be told when they're talking to an AI, and that AI-generated content needs to be labeled.

What surprised me: almost every "compliance" tool I found was a frontend banner generator. No audit trail, no proof the disclosure was actually shown, nothing tied to the real API call. If a regulator ever asked "can you prove this user was told?", most setups couldn't answer that.

So I built Disclarion an SDK that sits between your app and your LLM provider (OpenAI, Anthropic, Gemini), logs every disclosure event against the actual model call, and adds real watermarks/labels to AI-generated content instead of a static badge.

Launching it on Product Hunt tomorrow. Still a solo build, so genuinely curious what this community thinks especially if you've dealt with AI Act compliance yourselves, or if there's an obvious gap I'm missing. Happy to answer anything about the technical side too.

If you want to poke around: https://www.disclarion.com feedback (especially "this wouldn't work because X") is more useful to me right now than upvotes.