AI Safety Bill Vetoed in California

^{Photograph: Creative Commons Attribution-Share Alike 4.0}

Governor Gavin Newsom of California has vetoed an artificial intelligence (AI) safety bill, SB 1047, after intense debate over whether the bill would cause California's AI boom to suffer or offer valuable protection against the dangers of advanced AI systems. 

The bill aimed to regulate the development of powerful AI systems by making the companies that develop them liable for implementing sufficient safety protocols to prevent “critical harms”. One example of “critical harm” given in the bill was an AI model being used to create a weapon that results in mass casualties.

The bill's author, State Senator Scott Wiener, claimed that some institutions spread an unprecedented level of misinformation about SB 1047. He specifically criticised executives from Y Combinator and Andreessen Horowitz for propagating the idea that the bill could send startup founders to jail. Although not entirely false, this could only have theoretically happened if a developer had committed perjury by lying on the AI safety reports that SB 1047 would have required.

Wiener also refuted claims that SB 1047 would drive AI startups out of California, asserting that it would have affected companies nationwide equally as long as they conducted business in California. He expressed disappointment with respected figures in the AI community, such as Dr. Fei-Fei Li, who wrote an opinion piece on the matter for Fortune magazine, sharing what Weiner perceived as inaccurate statements about the bill's impact on open-source programmes. 

Despite the veto, Wiener believes that SB 1047 has significantly elevated the conversation around AI safety in California. Governor Newson declared in his letter rejecting the bill that he agreed with Wiener on the importance of taking action to protect the public before a major catastrophe takes place, and that safety protocols must be adopted, but that he doesn’t agree with settling “for a solution that is not informed by an empirical trajectory analysis of Al systems and capabilities”. 

In response to the debate, Governor Newsom has announced a new task force to develop guidelines for responsible AI development, and the state has passed 18 other AI-related laws in September.

Wiener hasn't ruled out the possibility of introducing a revised version of SB 1047 in the future, emphasising his continued commitment to promoting AI safety.