California Eyes New AI Guardrails with SB 53

A new California bill, SB 53, seeks to establish AI safety guardrails by compelling major labs to publish risk reports. The revised legislation, championed by Senator Scott Wiener, aims to balance innovation with critical oversight.

California State Senator Scott Wiener is a man intimately familiar with the formidable currents of Silicon Valley, a force that can both propel groundbreaking innovation and, as he’d argue, obstruct essential public safeguards.

After a bruising legislative defeat last year, Wiener has returned to Governor Gavin Newsom’s desk with a refined, yet no less critical, artificial intelligence safety bill, SB 53.

This time, the political winds appear to be shifting, offering a glimmer of hope that the Golden State might finally cement some of the nation’s first significant guardrails on an industry often accused of moving too fast and breaking too many things.

His first attempt, SB 1047, met a fiery end in 2024.

Tech giants, fearing the specter of liability for potential AI harms, waged a fierce campaign, arguing the bill would stifle America’s burgeoning AI boom.

Governor Newsom, echoing these concerns, ultimately wielded his veto pen, much to the delight of a popular AI hacker house that promptly celebrated with an “SB 1047 Veto Party,” where one attendee gleefully declared, “Thank god, AI is still legal.”

The episode underscored the immense power and influence wielded by an industry accustomed to operating largely beyond the reach of comprehensive regulation.

But Wiener, representing San Francisco, the very crucible of AI innovation, is not easily deterred.

His new bill, SB 53, represents a strategic recalibration.

Gone is the controversial liability clause that drew so much ire.

In its place is a focus on transparency and reporting.

If signed into law, SB 53 would compel leading AI laboratories—specifically those raking in over $500 million in revenue—to publish safety reports for their most capable models.

These reports would detail how their systems could contribute to the gravest of risks: human deaths, massive cyberattacks, and the creation of chemical weapons.

Currently, such disclosures are voluntary and often inconsistent, leaving a significant blind spot in understanding the potential societal impact of these powerful technologies.

The shift in approach appears to be paying dividends.

Where SB 1047 faced a united front of opposition, SB 53 has garnered crucial industry endorsements.

Anthropic, a prominent AI developer, has thrown its weight behind the bill.

Meta, another tech behemoth, expressed support for regulation that balances innovation with guardrails, calling SB 53 “a step in that direction.”

Even a former White House AI policy advisor, Dean Ball, sees it as a “victory for reasonable voices,” predicting a strong chance of Newsom’s signature.

This newfound, albeit cautious, acceptance from some corners of Silicon Valley suggests a growing recognition that some level of oversight is inevitable, and perhaps, even desirable.

Yet, not everyone is entirely onboard.

OpenAI, in a letter to Governor Newsom, argued that AI labs should only be beholden to federal standards – a curious stance to take with a state governor.

Venture capital firm Andreessen Horowitz has also vaguely hinted that California’s state-level efforts might run afoul of the Constitution’s dormant Commerce Clause, which prevents states from unfairly limiting interstate commerce.

These arguments betray a lingering desire for a singular, less stringent regulatory framework, ideally one crafted at a distance from the progressive impulses of California.

Senator Wiener, however, remains resolute in his conviction that states must lead.

He openly expresses a profound lack of faith in the federal government’s ability to enact meaningful AI safety regulation, particularly under a potential Trump administration.

He views recent federal efforts to preempt state AI laws as a form of “Trump rewarding his funders,” lamenting a perceived shift from the Biden administration’s emphasis on AI safety to a singular focus on growth.

Indeed, Vice President J.D. Vance’s remarks at a Paris AI conference – “I’m not here this morning to talk about AI safety… I’m here to talk about AI opportunity” – perfectly encapsulate this perceived pivot, a stance enthusiastically applauded by an industry that regularly sees its CEOs dining at the White House.

Wiener’s frustration is palpable.

“Every time I see tech CEOs having dinner at the White House with the aspiring fascist dictator, I have to take a deep breath,” he shared, his voice betraying a mix of exasperation and concern.

He is quick to clarify that he is not “anti-tech” and desires innovation, acknowledging the brilliance and wealth generated by these companies.

However, he firmly believes that “this is an industry that we should not trust to regulate itself or make voluntary commitments.”

It’s a stark reminder that even in the heart of capitalism, the public interest demands a legislative hand, especially when the potential harms are as catastrophic as those posed by unchecked AI.

Beyond the reporting requirements, SB 53 also establishes protected channels for AI lab employees to report safety concerns to government officials, a crucial whistleblower mechanism.

It also creates CalCompute, a state-operated cloud computing cluster designed to democratize AI research resources, moving beyond the exclusive domain of Big Tech.

These provisions reflect a broader strategy to foster both transparency and a more responsible, distributed ecosystem for AI development.

For Senator Wiener, this journey has been a “roller coaster,” yet profoundly “rewarding.”

He sees his efforts as instrumental in elevating the conversation around AI safety, not just in California, but globally.

His message to Governor Newsom is clear and concise: “We heard you.”

The new bill, he argues, meticulously follows the path laid out by the Governor’s previous veto message and the subsequent working group’s recommendations.

It is a testament to the legislative process, albeit a contentious one, attempting to thread the needle between fostering innovation and safeguarding humanity from the most severe, even existential, risks of a technology that is rapidly reshaping our world.

The coming weeks will reveal whether Governor Newsom agrees that the path has been successfully navigated, and whether California will indeed lead the charge in bringing accountability to the AI frontier.