California AI Safety Bill Vetoed: Governor Newsom Rejects Regulation of Massive AI Models

On Sunday, California Governor Gavin Newsom vetoed a historic bill that sought to create the nation’s first safety protocols for massive AI models.

The judgment is a serious setback to attempts to control the domestic business, which is multiplying and receiving minimal regulation. According to those who supported the measure, it would have created some of the first national controls on large-scale AI models and opened the door for national AI safety laws.

In response to congressional delay, the Democratic governor stated earlier this month to an audience at Dreamforce, an annual conference organized by software behemoth Salesforce, that California must take the lead in regulating AI, but that the proposal “may have a chilling effect on the industry.”

Also Read: California Approves AI Laws: Safeguarding Workers, Regulating Deepfakes, and AI Model Safety

The proposal, which faced strong resistance from tech giants, startups, and some Democratic House members, might have harmed domestic business by imposing onerous regulations, according to Newsom.

Rather, on Sunday, Newsom declared that the state will collaborate with some business leaders, among them AI pioneer Fei-Fei Li, to create safeguards around potent AI models. Li was against the AI safety plan.

To prevent the models from being manipulated to, for example, destroy the state’s power infrastructure or assist in the development of chemical weapons, the policy, which was intended to reduce possible hazards produced by AI, would have compelled corporations to test their models and publicly reveal their safety protocols. Experts predict that given how quickly the business is developing, those outcomes may come to pass in the future. Additionally, workers would have been protected as whistleblowers.

The bill is one of many that the legislature has enacted this year to control AI, combat deepfakes, and safeguard employees. Legislators from the state emphasized that California needs to act this year, emphasizing the painful lessons they had to learn from not regulating social media giants when they had the opportunity.

Also Read: Clearview AI Settles Privacy Lawsuit with 23% Stake for Affected Americans

Proponents of the concept, such as Anthropic and Elon Musk, said that it may have brought some degree of accountability and transparency to large-scale AI models, as experts and developers claim they still don’t fully understand the behavior and motivations of these models.

The bill specifically targeted technologies that would cost over $100 million to develop. Although none of the AI models in use today have reached that point, some experts predict that may change in the coming year.

“This can be attributed to the significant increase in investment within the industry,” stated Daniel Kokotajlo, a former researcher at OpenAI who resigned in April due to his perception of the company’s indifference to AI concerns. “This is a crazy amount of power to have any private company control unaccountably, and it’s also incredibly risky.”

The US is already lagging behind Europe in terms of risk-reduction regulations for AI. Though backers of the California proposal said it would have been a good first step to create guardrails around the rapidly developing technology that is prompting concerns about job loss, misinformation, breaches of privacy, and automated bias, the measure wasn’t as extensive as those in Europe.

Also Read: Colorado Enacts Historic AI Discrimination Law to Protect Workers and Consumers

Last year, several of the top AI firms freely consented to abide by the White House’s safety measures, which included testing and disclosing information on their models. According to the bill’s proponents in California, AI developers would have been required to adhere to standards akin to those pledges.

Nevertheless, some said that the bill would “kill California tech” and impede innovation, among them former Speaker of the US House Nancy Pelosi. They said that it would have deterred AI developers from making substantial model investments or disclosing open-source software.

Big tech corporations and AI developers in California have won again thanks to Governor Gavin Newsom’s decision to reject the measure. These groups have been working with the California Chamber of Commerce to influence lawmakers and the governor over the past year to stop the advancement of AI legislation.

Two more comprehensive AI initiatives that also encountered growing resistance from the tech sector and other quarters perished last month, just before a parliamentary deadline. The bills would have prohibited discrimination from AI tools used to make hiring choices and mandated that AI developers label content generated by AI.

Also Read: World First AI Act Passed by European Parliament; New Laws for Artificial Intelligence Restrictions