The draft European AI Act
The European Commission proposed the world's first comprehensive AI law, sorting applications by level of risk and banning some outright.
Why it matters
Regulation ran ahead of the market for the first time: the law was written about systems that did not yet exist at scale.
The scheme is built on risk tiers: prohibited practices, high risk with mandatory requirements, limited transparency obligations, and free use. Social scoring and certain kinds of biometric surveillance fall under the prohibitions. The proposal appeared eighteen months before ChatGPT, so the provisions on general-purpose models had to be written in during the negotiations.