AI companies operating in California are grappling with tricky compliance questions as they navigate the state’s first-in-the-nation law that mandates disclosures of their safety policies, such as how to protect their trade secrets.
The law (SB 53) is a negotiated-down version of a previous bill that would have imposed liability on developers for catastrophic risks. Compliance with the new version will be an easier lift, said Jessica Ji, senior research analyst at Georgetown University’s Center for Security and Emerging Technology.
“The companies that SB 53 would apply to are basically already doing these things. They are all already ...