The European AI Act enters into force
The first general law on artificial intelligence anywhere entered into force in the European Union, with obligations graded by level of risk and fines of up to 7 per cent of worldwide turnover.
Why it matters
Regulating AI stopped being an intention and became law in force, with dates, duties and penalties.
The act divides applications into levels: forbidden, such as state scoring of people or emotion recognition at work; high risk, such as medicine, hiring, borders and justice, with a mandatory conformity assessment; limited, with a duty to tell a person they are talking to a machine. A separate chapter covers general-purpose models: documentation, respect for copyright, and above a given amount of training compute a systemic-risk assessment as well. The obligations phase in over three years, so 1 August started a countdown rather than enforcement.