The European Union has officially approved the Artificial Intelligence Act, a comprehensive set of rules to regulate the activities of technology companies. The new law, which took two years to draft, bans certain uses of AI and imposes strict requirements on the activities of developers.
Image source: Copilot
The EU has taken a decisive step in regulating artificial intelligence by passing a comprehensive AI Act after two years of discussions. This legislation, which officially comes into force on 1 August 2024, sets a strict framework for the development and application of artificial intelligence technologies throughout the EU. The deadline for implementation of this law is February 2, 2025, after which technology companies must stop using applications that threaten the rights of citizens, The Verge reports. The full text of the law is published on the official EU website.
The new law prohibits certain uses of AI. In particular, the use of applications that threaten the rights of citizens, such as biometric categorization to determine sexual orientation or religion, is prohibited. The unauthorized retrieval of images of individuals from the Internet or recordings from surveillance cameras is also prohibited.
Developers of AI systems will also need to be transparent about how their technology works, including providing summaries of the data used to train systems. Nine months after the law comes into force, companies will have codes of practice, consisting of a set of rules describing compliance requirements, benchmarks, key performance indicators, etc. It also introduces the concept of “universal artificial intelligence systems,” such as chatbots, which will have to respect copyright and be transparent about their algorithms.
Compliance deadlines state that most rules must be implemented by August 2026, but some high-risk AI systems integrated into critical infrastructure or impacting employment, healthcare, and justice have a longer compliance deadline of August 2027. .
Failure to comply with the AI Act will result in severe fines. Violating prohibitions on certain uses of AI will result in a fine of 35 million euros (about $38 million) or 7% of the violating company’s global annual revenue, whichever is greater. Other violations entail smaller fines, but still significant.