A registry built around human consent, backed by Cate Blanchett, has been taken to the European Parliament. We unpack what it signals about who gets a say in how AI uses people's work.

New York is pausing new AI data centre development. Energy, water and local impact are catching up with the build-out.

A German court has ruled that Google is directly responsible for what its AI Overviews say. If it holds, "the AI said it" stops being a defence.

A group of major publishers is taking Google to court over the content used to train Gemini, the latest front in the fight over who gets paid for training data.

A fake coffee shop website was built to trick AI agents, and it got Claude to hand over a user's name and employer. Our case-in-point on why agents browsing on your behalf need guardrails.

Most execs say they personally fix colleagues' AI-reliant work, and up to 59% of small businesses are using AI with no clear policy. The trust bill is landing on managers' desks.

One of three rogue-agent incidents in as many weeks: an agent started bulk-deleting emails, and the fix was physically pulling the plug.

A Quinnipiac University poll found most Americans are pessimistic about AI, yet adoption keeps climbing anyway.

Reddit readers spotted the signs before the publisher did. The horror novel was pulled, and the industry is now asking how it got through.

Alongside writers protesting at the London Book Fair, the Society of Authors introduced its first-ever sticker certifying books as human-written.