California’s new AI safety law shows regulation and innovation don’t have to clash | TechCrunch
Key Takeaways
- California Governor Gavin Newsom signed SB 53, a first-in-the-nation bill requiring large AI labs to disclose safety and security protocols.
- The law specifically targets preventing catastrophic risks, such as the use of AI models for cyberattacks or bioweapon development.
- Adam Billen of Encode AI supports the bill, arguing it enforces existing safety standards that companies might drop under competitive pressure.
- The AI industry generally opposes state regulation, fearing it hinders progress against international competitors like China, leading to significant lobbying efforts.
- Advocates are actively fighting proposed federal legislation, like the SANDBOX Act, which seeks to preempt or override state-level AI regulations.
California Governor Gavin Newsom recently signed SB 53, an AI safety and transparency bill, which mandates that large AI labs disclose their protocols for preventing catastrophic risks like cyberattacks or bioweapon creation, with enforcement by the Office of Emergency Services. Adam Billen, VP of Public Policy at Encode AI, praised the bill, arguing that it enforces existing safety promises that companies might otherwise abandon under competitive pressure, such as when rivals relax standards. Silicon Valley and many AI labs oppose such regulation, fearing it will hinder the U.S. in its technological race against China, leading powerful entities to fund pro-AI politicians and push for federal moratoriums. Billen confirmed that Encode AI successfully fought a previous moratorium proposal and is now watching for new federal preemption strategies, like Senator Cruz's SANDBOX Act, which could override state laws. He warned that narrowly scoped federal legislation could effectively eliminate federalism for critical technology regulation. While acknowledging the importance of the race with China, Billen asserts that killing state bills focusing on issues like deepfakes and algorithmic discrimination is not the correct approach to maintaining U.S. progress.
