NEW EU AI RULES ARE LIVE. READ THIS BEFORE YOU SIGN OFF ON THAT AI SLOP.
On 2 August 2026, new EU transparency rules for AI-generated content came into force. Here's what changed, who it affects, and why craft still wins.
The date that matters: 2 August 2026
Article 50 of the EU AI Act is now live. If your brand serves EU audiences, this applies to you — regardless of where you're based. UK and Irish studios and brands are firmly in scope.
The four things you now have to disclose
- Chatbots must say they're AI. No pretending to be a human on the other end.
- AI-generated or altered images, audio, video and text need a machine-readable mark. Synthetic content has to be detectable as synthetic.
- Deepfakes and AI-written text on public-interest topics need clear, visible labels. Not buried in a footer.
- Emotion-recognition and biometric AI systems must tell people they're being read. Consent, not surprise.
What's exempt
Minor edits — grammar fixes, colour correction, standard retouching — don't count as "AI-generated." And content made and published before 2 August 2026 doesn't need retroactive labelling.
The penalty for getting it wrong
Up to €15 million, or 3% of global annual turnover — whichever's higher.
Where we stand
We already sit on the right side of this one. AI has a role in our process — as an accelerant, not a shortcut — but every piece of work carries a human hand on the final call. Compliance isn't a redesign for us. It's just how we already work.
Where this goes next
Audit your content pipeline now. Know which assets will need marking, and build the habit of disclosure before enforcement catches up with anyone caught out. If you want help figuring out what actually applies to your brand, we're happy to talk it through.
Want to talk it through?