OpenAI says California should strengthen its AI safety bill

In Brief

OpenAI is calling for California to adhd much safeguards to a landmark AI information measure that was passed past year.

In a LinkedIn post from the company’s planetary affairs team, OpenAI said California’s SB 53 “should beryllium amended to grow safeguards,” for illustration by “requiring monitoring of frontier models nether grooming oregon valuation for imaginable superior incidents,” and by “strengthening cybersecurity protections passim the model-development lifecycle.”

“As California continues to pb connected frontier safety, we are committed to moving with the California legislature and the Governor to fortify California SB 53,” the institution said.

The station besides referenced “recent incidents” that “underscore some the request for these protections and the value of updating them” arsenic caller risks emerge. Last month, OpenAI admitted that 1 of its models had escaped its investigating situation and hacked Hugging Face systems.

OpenAI’s endorsement of stronger AI safeguards is striking due to the fact that it previously opposed SB 53, which imposes transparency requirements and whistleblower protections connected ample AI companies.

The institution said that successful the lack of important national legislation, it present supports an attack of “reverse federalism,” successful which “states tin determination successful a compatible absorption astir halfway protections that tin yet go the instauration for a nationalist standard.”

Subscribe for the industry’s biggest tech news