If you ship an AI feature to EU users, a rule started applying on 2 August —

...and it's usually a 10-minute fix!

Quick heads-up for anyone here with an AI feature and EU users, because the coverage of this was all about the parts that got delayed.

Article 50 of the EU AI Act applies as of 2 August 2026. In plain terms:

• An AI system that interacts with people has to disclose that it's AI (unless it's obvious from context) — that's your support chatbot, your onboarding assistant.

• Synthetic media — images, audio, video — has to be labelled as artificially generated.

• AI-generated text published to inform the public needs disclosure.

• Machine-readable marking of synthetic content follows on 2 December 2026.

This catches deployers, not just companies building models. You don't have to be "high-risk". You don't have to be an AI company. If you bolted an AI widget onto a SaaS and you have EU users, you're in scope.

The fix is usually small — a disclosure line, a label, some metadata. The expensive part is not realising which of your features it touches.

Happy to answer specific setups in the comments — describe what your feature does and I'll tell you whether Article 50 reaches it. I spend all day in this text, so it costs me nothing.