Lawmakers and startups lock horns over California AI legislation

California republic flag pictured in San Francisco with blue skies in background.
(Image credit: Getty Images)

Dozens of California-based tech companies have hit out at proposed AI regulations and oversight requirements in the state in an open letter addressed to lawmakers.

Penned by venture capital firm Y Combinator, the letter’s signatories include a list of over 140 tech companies pursuing AI development in the state, including firms working in areas spanning healthcare and finance. 

The letter takes issue with Senate Bill 1047, legislation devised by Senator Scott Wiener, that would impose AI safety testing measures on AI models. 

The group represented by Y Combinator thinks this bill could “inadvertently threaten the vibrancy of California's technology economy and undermine competition”.

While the companies acknowledged the state’s commitment to “AI's safe and ethical development” and appreciate the amendments so far made to the bill to reflect stakeholder concerns, their “core concerns” about the bill remain. 

The group laid out four key concerns. In the first instance, Y Combinator criticized the level of burden placed on businesses by the legislation and argued that AI developers themselves should not be made liable for acts of AI misuse.

They then criticized the “problematic” regulatory threshold imposed by the bill, as AI development is still evolving and thus “specific metrics” may quickly become arbitrary as future models are rolled out. 

As non-Californian companies will not face such regulatory thresholds, the letter noted, AI developers could be liable to move out of the state, negatively impacting the region’s market. 

The group also took issue with the legislation’s imposition of a “killswitch” for AI models. They argued that such a function could act as a “de facto ban on open-source AI development”.

Finally, they argued that the terms of the bill were vague. So vague, they claimed, that the legislation’s terms could prove “fodder for judges to interpret however they like” and potentially lead to more restrictive laws.

Simon Bain, CEO and founder at San Jose-based firm OmniIndex, echoed the sentiments of the letter in comment to ITPro, though the firm has not signed the complaint itself.

"Regulators need to stop wasting their time focussing on scaremongering and kill switches and start looking at the real issues of LLMs today,” Bain said. 

He argued that regulators should be focusing on the more pertinent issues of AI, such as the potential environmental damage and the ongoing issues of copyright infringement, for example.

"By focussing on these issues, California could be a true leader in the future of AI,” Bain said. “Instead, the state finds itself at war with the innovators over grandstanding on tomorrow's issues".

Debrup Ghosh, manager at California-based Synopsys, was slightly more lenient in comment to ITPro. He acknowledged that the construction of legislative guardrails was of paramount importance in California.

That being said, Ghosh accepted that, in order to foster innovation and retain talent in the state, policy makers “need to work with diverse stakeholders to ensure regulations don’t impede economic growth and innovation”.

California firms are the latest to push back against AI regulation 

Rising tensions in California bear a resemblance to that witnessed in Europe over the introduction of the EU AI Act. Industry stakeholders repeatedly voiced concerns over the potential impact of the legislation on AI development in the regent.  

In November last year, a joint statement from tech policy group, DigitalEurope, warned that startups in the AI space could be “regulated out of existence” by the legislation.  

DigitalEurope claimed the EU AI Act limited organizations' ability to utilize foundation models and it argued that the legislation could harm long-term innovation.

According to Gregor Hofer CEO of Rapport and Speech Graphics, there are parallels to be drawn between the EU AI Act and the bill in California.


“The reaction to California’s proposed AI legislation mirrors the backlash against the EU AI Act, with stakeholders in both regions voicing similar concerns about potential negative impacts on innovation and competitiveness,” Hofer said.

“Strict AI legislation in California could deter start-ups and established businesses from conducting cutting-edge research and development in the state by imposing significant compliance costs. The knock on effect is that this would likely hamper innovation and competitiveness,” he added. 

Stakeholders from the European open source community also voiced concerns over the EU AI Act, warning that it may negatively impact the space. 

George Fitzmaurice
Staff Writer

George Fitzmaurice is a staff writer at ITPro, ChannelPro, and CloudPro, with a particular interest in AI regulation, data legislation, and market development. After graduating from the University of Oxford with a degree in English Language and Literature, he undertook an internship at the New Statesman before starting at ITPro. Outside of the office, George is both an aspiring musician and an avid reader.