Legislators in California voted on Tuesday to move forward with a bill that would mandate artificial intelligence companies to test their systems and implement safety measures to prevent potential misuse, such as causing a blackout of the state’s electric grid or aiding in the creation of chemical weapons. Experts warn that these scenarios could become a reality as technology advances rapidly.
The proposed legislation, the first of its kind, aims to mitigate the risks associated with AI. However, it is facing strong opposition from venture capital firms and tech giants like Meta (the parent company of Facebook and Instagram) and Google, who argue that the regulations should target the end-users rather than developers.
Sponsored by Democratic state Sen. Scott Wiener, the bill seeks to establish safety standards to prevent the emergence of highly powerful AI models that could pose significant risks. The regulations would only apply to systems that require more than $100 million in computing power for training, a threshold that no current AI models have reached.
During a legislative hearing, Wiener criticized the misinformation spread by opponents of the bill and clarified that it does not introduce new criminal charges for developers if they have tested their systems and taken precautions to mitigate potential risks. Legal actions for violations would be pursued by the state attorney general under the proposed law.
While Gov. Gavin Newsom has championed California’s role as a pioneer in AI adoption and regulation, he has not commented on the bill specifically. Tech companies fear that the proposed requirements could impede innovation and hinder the development of large AI systems.
Proponents of the bill, including prominent AI researchers, argue that California cannot afford to delay regulation, drawing parallels to the negative repercussions of delayed action against irresponsible social media practices. The legislation also seeks to establish a new state agency to oversee developers and promote industry best practices.
Additionally, legislators are considering two other measures aimed at safeguarding Californians from potential AI-related harms, addressing issues like automation discrimination in hiring and renting practices, as well as restricting social media companies from collecting and selling data of minors without consent.
Related story: Could your vote be swayed by artificial intelligence this November?
