EU Sets New Rules to Ensure AI is Safe and Ethical
-
The EU's Artificial Intelligence Act aims to establish world-leading rules to ensure AI is human-centric and under human control. High risk AI uses face stricter requirements.
-
The law bans certain high risk AI uses like social scoring systems and some predictive policing. There are also bans on using AI for remote biometric identification in public spaces.
-
New provisions address emerging generative AI models like chatbots. Developers must label manipulated content, assess risks, and report incidents. Extra rules apply to powerful models like GPT4.
-
The AI Act is expected to formally become EU law by mid-2023. Provisions will phase in over time, with complete regulations in place by mid-2026.
-
Each EU country will have an AI regulator to enforce rules. Violations could draw fines up to 35 million euros or 7% of global revenues.