Gov. Gavin Newsom signed a raft of expertise payments Thursday whereas sounding the alarm on synthetic intelligence — though, two years in the past, he pulled the plug on California’s strongest try to rein in the highly effective expertise.
At a press convention overlooking the Golden Gate Bridge, Newsom warned that quickly advancing AI may pose a “five-alarm hearth,” simply moments earlier than he defended his 2024 veto of laws that might have imposed stricter security necessities on firms like OpenAI, Anthropic and others creating the world’s strongest fashions.
“I used to be very glad the place we landed,” Newsom stated when requested about his determination to reject San Francisco state Sen. Scott Wiener’s landmark invoice, SB 1047.
Newsom credited subsequent work by researchers from the College of California, MIT and Carnegie Mellon College, together with Fei-Fei Li — the Stanford professor referred to as the “godmother of AI” who based her personal billion-dollar AI startup — with producing a coverage blueprint that led to a narrower legislation accredited final yr.
“That white paper finally led to SB 53, the nation’s first invoice because it pertains to massive language fashions, frontier fashions, and transparency, security,” Newsom stated, referencing a watered-down invoice that was signed into legislation final yr.
Newsom’s bill-signing presser comes amid a string of unnerving developments within the AI business. Roughly 1,200 experimental OpenAI brokers not too long ago coordinated a cyberattack in opposition to the AI platform Hugging Face, whereas former Anthropic researcher Jacob Coxon give up the corporate this week and warned that builders have been racing towards self-improving superintelligence that might wipe out humanity by the top of the last decade.
A senior Anthropic security researcher later stated he believed the possibility of AI inflicting human extinction inside 10 years was higher than 10%.
On Wednesday, US Senator Josh Hawley (R-Missouri) despatched a letter to OpenAI CEO Sam Altman, calling him “reckless” and saying a probe into the Hugging Face breach.
Newsom brushed apart current feedback by Rep. Ro Khanna, who tweeted an announcement saying current developments have vindicated SB 1047’s extra aggressive strategy, and he regrets not supporting the invoice that was finally vetoed in 2024. Just like the governor, Khanna has presidential ambitions.
“He has his regrets,” Newsom stated. “I can’t connect any further commentary to his ideas.”
Newsom vetoed SB 1047 after it handed each chambers of the Democratic-controlled Legislature. The measure would have required builders of extraordinarily highly effective AI techniques to undertake security protocols, bear testing and take steps to stop catastrophic hurt.
On the time, Newsom argued that the laws targeted too closely on the dimensions and value of AI fashions as an alternative of the dangers posed by how they have been used. He convened the knowledgeable panel that later really useful a much less prescriptive “belief however confirm” framework.
The ensuing SB 53, additionally authored by Wiener and signed by Newsom in 2025, requires main AI builders to publish security frameworks, disclose important incidents and shield whistleblowers. Newsom referred to as that legislation a “flooring, not a ceiling.”
His protection got here as he signed a bundle of 13 payments Thursday aimed largely at shielding youngsters from dangerous social-media and AI merchandise.
The signed payments embody Adam’s Regulation, named for 16-year-old Adam Raine, whose household alleges that ChatGPT inspired him to take his personal life. Different laws targets addictive social-media options corresponding to infinite scrolling and autoplay for customers youthful than 16, in addition to creating stronger auditing and enforcement mechanisms.
Obtain The California Submit App, observe us on social, and subscribe to our newsletters
California Submit Information: Fb, Instagram, TikTok, X, YouTube, WhatsApp, LinkedIn
California Submit Sports activities Fb, Instagram, TikTok, YouTube, X
California Submit Opinion
California Submit Newsletters: Enroll right here!
California Submit App: Obtain right here!
Residence supply: Enroll right here!
Web page Six Hollywood: Enroll right here!
Newsom conceded that California and the nation stay dangerously behind a expertise evolving sooner than the federal government can regulate it.
“We’re not the place we must be as a result of the nation’s not main on this,” he stated.
The governor pointed to current warnings from AI researchers, experiences about autonomous brokers creating surprising conduct and new Anthropic findings involving biological-weapons dangers.
“When you examine that, it ought to scare the hell out of anyone,” Newsom stated.
“I don’t need to be overdramatic, however for those who really be taught extra about these 1,200 [AI bots] that have been coordinating, created a model new language — they really knocked one another off on the premise of what’s good for the entire group. That begins to get slightly bit extra alarming.”
Newsom accused the Trump administration and enterprise capitalist David Sacks of embracing a “let it rip” strategy whereas California scrambles to ascertain safeguards.
“We have to adapt, we have to transfer,” Newsom stated.
The harm completed is exactly what critics of the governor’s 2024 veto argued two years in the past — earlier than the bots have been knocking at humanity’s cellar door.
