AI Systems With 'Unacceptable Risk' Are Now Banned In the EU

The European Union has introduced regulations to ban AI systems that pose "unacceptable risk" or harm, including activities such as social scoring, deceptive manipulation, and unauthorized facial recognition database expansion. These regulations categorize AI applications into four risk levels: minimal, limited, high, and unacceptable. Minimal risk applications, like email spam filters, will not face regulatory oversight, while limited risk applications, such as customer service chatbots, will have light oversight. High-risk applications, like AI for healthcare recommendations, will undergo heavy regulatory scrutiny, and unacceptable risk applications will be prohibited entirely.

Companies using banned AI applications in the EU will face fines of up to ~$36 million or 7% of their annual revenue, whichever is greater, regardless of their headquarters' location. Compliance is expected by February 2, with enforcement provisions and identification of competent authorities set to take effect by August. This regulatory framework aims to mitigate risks associated with AI technologies and ensure ethical use within the EU.

Key takeaways

The European Union can now ban AI systems that pose "unacceptable risk" or harm.
Unacceptable AI activities include social scoring, deceptive manipulation, and unauthorized facial recognition database expansion.
AI applications are categorized into four risk levels: minimal, limited, high, and unacceptable risk.
Companies using prohibited AI applications in the EU may face fines up to ~$36 million or 7% of annual revenue, with full compliance expected by February 2.

AI Systems With 'Unacceptable Risk' Are Now Banned In the EU - Slashdot

Key takeaways

Discussion (0)