Every The AI Downside article filed under AI Regulation — the failures, dark patterns, price hikes and privacy grabs we've logged here, newest first. 5 and counting.
A plain-English guide to the law on AI deepfakes of your face and voice in 2026 — the right of publicity, the TAKE IT DOWN Act, the NO FAKES Act, US state laws, Denmark's copyright approach and the EU AI Act — and where the gaps still leave you exposed.
An AI tool tells you something wrong, you act on it, you’re out of pocket — or worse. Who pays? A tour of the terms you clicked past, the cases that pierced them, and the law that is still being written.
When an AI tool retires a feature or you want to leave, can you take your data and creations with you? Data-protection law gives you a portability right — but it covers data you provided, not data the model inferred or generated, and it rarely forces a working export button. Here's the mechanism, the law, and the gap in between.
The GDPR gives you the right to have your data erased. But a trained model stores your data as distributed weights, not deletable records — so “delete” becomes retraining or approximate “unlearning” that forgets imperfectly. Here’s the mechanism, and why it matters.
AI regulation is real but uneven — and does less for the individual user than the headlines suggest. The EU’s toughest safeguards just slipped to 2027, the rules that did land are softer than they sound, and the US is trying to undo its own. A plain-English map.
