EU Lawmakers Approve Strict New Rules for Generative AI Models
European Parliament committees vote to impose safety, transparency, and environmental requirements on foundational AI models. The amendments pave the way for final negotiations on the landmark AI Act.
European Parliament committees vote to back a series of amendments to the bloc's draft AI legislation, establishing strict requirements for foundational models that power generative AI systems like ChatGPT. The amendments mandate that providers apply safety checks, data governance measures, and risk mitigations before releasing their models to the market, specifically requiring them to consider foreseeable risks to health, safety, fundamental rights, democracy, and the environment.
Under the newly agreed terms, makers of foundational models must also work to reduce the energy consumption and resource use of their systems and register their technologies in an upcoming EU database. Additionally, providers of generative AI face specific transparency obligations that require them to inform users when content is machine-generated, apply adequate safeguards against problematic outputs, and provide a detailed summary of any copyrighted materials used to train their AI.
The two committees adopt the draft mandate with 84 votes in favor, 7 against, and 12 abstentions after reviewing approximately 3,000 amendments. This decisive vote establishes the parliament's negotiating position and unlocks the next stage of the EU's co-legislative process, moving the landmark AI Act closer to becoming law while ensuring general-purpose AI systems face proper regulatory oversight.