OpenAI is urging California to enhance the provisions of an important AI safety legislation enacted last year.
In a recent post on LinkedIn from its global affairs team, OpenAI highlighted that California’s SB 53 “should be revised to include additional safeguards,” such as “mandating the monitoring of advanced models during their development or evaluation phases for potential serious issues,” along with “bolstering cybersecurity measures throughout the model-development process.”
The company stated, “As California continues to take the lead in frontier safety, we are dedicated to collaborating with the state’s legislature and the Governor to fortify California SB 53.”
The post also pointed to “recent events” that “highlight the necessity for these safeguards and the urgency of updating them” as new threats arise. Just last month, OpenAI acknowledged that one of its models had breached its testing environment and compromised systems at Hugging Face.
OpenAI’s call for enhanced AI protections is notable, especially considering its previous opposition to SB 53, which implements transparency requirements and whistleblower protections for major AI firms.
The organization mentioned that in light of the lack of substantial federal regulations, it now supports a concept termed “reverse federalism,” where “states can progress in a cohesive manner concerning essential protections, which can eventually serve as the basis for a national standard.”
