EU adopts draft AI law

 The draft categorises AI as either “unacceptable,” “high,” “limited” or “minimal.” 

Unacceptable examples are  AI that manipulates vulnerable groups of people, which leads to predictive policing systems and which employs biometric categorisation using gender, race, ethnicity and religion, among other identifiers.

High risk will be assessed before being put on the market and throughout its life-time. 

Limited risk AI will have to comply with transparency requirements, such as making users aware they are interacting with AI.

Generative AI like ChatGPT will have to disclose its role in content generation. It will be prevented from generating illegal content and will have to publish any copyrighted data used to train AI systems.