The California State Assembly has approved a bill that lays the groundwork for regulating artificial intelligence (AI) within the state. This landmark development, encapsulated in SB 1047, mandates that developers meet specific conditions before commencing AI model training activities. These prerequisites include safety protocols, shutdown capabilities, and cybersecurity protection measures. Senator Scott Weiner introduced the bill, which defines “covered models” based on machine learning metrics such as the quantity and cost of computing power used during training. The bill also requires developers to report any safety incidents.
Passed with a 49-15 vote, SB 1047 aims to minimize the potential misuse of AI technology to safeguard public safety. Notably, Elon Musk expressed his support for the bill on social media, while Nancy Pelosi criticized the legislation, stating it was “well-intentioned but ill-informed” and potentially harmful to American consumers. SB 1047 aligns with concerns raised in the White House’s Blueprint for an AI Bill of Rights, which addresses AI’s societal impacts.
Following the State Assembly’s approval, the bill returns to the Senate for further action. Should it pass in the Senate, it will proceed to Governor Gavin Newsom, who will decide whether to sign it into law. If enacted, California will join the ranks of other jurisdictions, such as the European Union, which has implemented AI legislation focusing on different risk levels associated with AI technologies.
For further details, the full text of the article is available on JURIST.