The bill comes amid a surge of state bills related to AI, with 407 AI-related bills currently active in 44 U.S. states. Several of these have already been signed into law. The rise in state-level bills could pressure Congress to pass federal AI legislation, as compliance with a single federal law may be easier for companies than navigating a multitude of different state laws. The proposed California law builds on an executive order by President Biden that requires safety tests on powerful AI systems. However, the California measure goes further, requiring hacking protections, protection for AI-related whistleblowers, and mandatory testing.
Key takeaways:
- A California state lawmaker, Sen. Scott Wiener, has introduced a bill that would require companies to test powerful AI models for unsafe behavior, implement hacking protections, and ensure the technology can be shut down completely before releasing them.
- The bill also mandates AI companies to disclose their testing protocols and safety measures to the California Department of Technology. If the AI technology causes critical harm, the state's attorney general can sue the company.
- The proposed California law builds on an executive order by President Biden, but goes further to require hacking protections, protect AI-related whistleblowers, and mandate companies to conduct testing.
- There are currently 407 AI-related bills active in 44 U.S. states, with several already signed into law. This could lead to greater industry pressure on Congress to pass AI legislation, as complying with a federal law may be easier than dealing with different state laws.