In a landmark move for artificial intelligence regulation, the California State Assembly has passed Bill SB 1047, setting the stage for new oversight on AI development.
The bill, approved with a 49-15 vote, introduces crucial safeguards for AI model training, reflecting a proactive approach to addressing public safety concerns associated with emerging technologies.
SB 1047, introduced by Senator Scott Weiner, outlines specific conditions that developers must meet before initiating the training of “covered models,” which are defined by the scale and cost of the computing resources used.
The legislation mandates that developers implement safety protocols, shutdown capabilities, and cybersecurity measures. Additionally, developers must report any safety incidents related to these models, aiming to mitigate potential risks associated with AI misuse.
Senator Weiner hailed the bill as a historic step towards safeguarding public interests while fostering technological advancement.
“With this vote, the Assembly has taken the truly historic step of working proactively to ensure an exciting new technology protects the public interest as it advances,” Weiner stated.
He stated that SB 1047 represents a sensible approach, reflecting the commitments that major AI companies have already made voluntarily.
The bill’s approval has garnered mixed reactions. Tesla CEO Elon Musk praised the legislation on X (formerly Twitter), identifying himself as a strong advocate for AI regulation.
In contrast, California Congresswoman Nancy Pelosi criticized the bill as “well-intentioned but ill-informed,” arguing that it could be more detrimental than beneficial in protecting consumers.
SB 1047 emerges in the wake of growing concerns about AI’s potential misuse, including the risk of contributing to the development of weapons of mass destruction.
The bill follows the White House’s Blueprint for an AI Bill of Rights released in 2022 and President Joe Biden’s 2023 executive order establishing some national AI safety standards.
Globally, the European Union has already implemented AI regulation, focusing on different risk levels associated with AI use, effective from August 1. The EU’s approach contrasts with California’s, which targets specific types of AI models.
With the Assembly’s passage, SB 1047 now awaits approval from the Senate. Following Senate deliberation, the bill will be presented to Governor Gavin Newsom for final consideration.
Should Newsom sign the bill into law, California will set a precedent in AI regulation, shaping the future of technological governance.
