FILE – California Gov. Gavin Newsom vetoed SB1046, a hotly contested measure that would have been the nation’s strictest AI safety law. (AP Photo/Rich Pedroncelli File)
Rich Pedroncelli/AP/FR171957 AP
hide caption
caption conversion
Rich Pedroncelli/AP/FR171957 AP
California Gov. Gavin Newsom on Sunday vetoed a bill that would have enacted the nation’s most sweeping regulations on the booming artificial intelligence industry.
California lawmakers overwhelmingly passed the bill, known as SB 1047, which is seen as a potential blueprint for national AI legislation.
The measure would hold tech companies legally responsible for harm caused by AI models. Additionally, the bill would require technology companies to enable a “kill switch” feature for AI technology in cases where the system is abused or fraudulent.
Newsom in a statement described the bill as “well-intentioned,” but noted that its requirements would require “stringent” regulations that could make life difficult for leading artificial intelligence companies. top of the state.
In his declareNewsom said the bill focuses too much on the biggest and most powerful AI models, and suggested that smaller upstarts could prove similarly disruptive.
“Smaller, specialized models could be as dangerous or even more dangerous than the models targeted by SB 1047 – at the potential cost of limiting the very innovation that drives progress,” Newsom wrote. beneficial to the public interest”.
California Sen. Scott Wiener, a co-author of the bill, criticized Newsom’s move, saying the veto was a step backward for artificial intelligence accountability.
“This veto brings us face to face with the disturbing reality that companies that want to create an incredibly powerful technology will face no binding constraints from policymakers,” Wiener wrote. US policies, especially as Congress continues to be paralyzed in regulating the tech industry in any meaningful way.” on X.
The now-rescinded bill would have forced industry to conduct safety tests on powerful AI models at scale. Without such requirements, Wiener wrote on Sunday, the industry would take policy into its own hands.
“While major AI labs have made admirable commitments to monitor and mitigate these risks, the truth is that voluntary commitments from industry are not enforceable and are rarely effective. works well for the public.”
Many powerful Silicon Valley companies, including venture capital firm Andreessen Horowitz, OpenAI and trade groups representing Google and Meta, have lobbied against the bill, arguing that it would slow development. development of AI and stifling the growth of early-stage companies.
“SB 1047 would threaten that growth, slowing the pace of innovation and causing California’s world-class engineers and entrepreneurs to leave the state to seek greater opportunities elsewhere,” Chief Strategy Officer by OpenAI Jason Kwon wrote in a letter sent to Wiener last month.
However, other tech leaders support the bill, including Elon Musk and pioneering AI scientists Geoffrey Hinton and Yoshua Bengio, who signed a letter urging Newsom to sign on to it.
“We believe that the most powerful AI models could soon pose serious risks, such as expanded access to biological weapons and cyberattacks on critical infrastructure. It is feasible and appropriate for pioneering AI companies to test whether the most powerful AI models can cause serious harm and for them to take appropriate safeguards against such risks. That’s the problem.” wrote Hinton and dozens of former and current employees of leading AI companies.
California Sen. Scott Wiener called the veto a “failure” in an X post for “all those who believe in oversight by large corporations making important decisions that affect to the safety and welfare of the public.”
Other states, like Colorado And Utahenacted more narrowly tailored laws to address how AI can perpetuate bias in employment and health care decisions, as well as other related consumer protection concerns to AI.
Newsom recently signed other AI bills into law, including one aimed at cracking down on the spread of deepfakes in elections. Another protects actors from having their likenesses copied by AI without their consent.
As billions of dollars pour into the development of AI and as it penetrates many corners of daily life, lawmakers in Washington have yet to propose a single federal law to protect people from its potential harmful effects, as well as failure to monitor its rapid development. .
