The tech community in California is split over a new bill that sets strict safety guidelines and a ‘kill switch’ for AI models, with some praising it for protecting against potential risks and others criticising it for stifling innovation and imposing liabilities.
Silicon Valley Reacts to Californian AI Safety Bill
California’s AI community is voicing strong opposition to a state bill that mandates strict safety measures and a “kill switch” for powerful AI models. The legislation, known as the Safe and Secure Innovation for Frontier Artificial Intelligence Systems Act, was passed by the state Senate last month and is set for a general assembly vote in August.
The bill requires AI developers in California—including major companies like OpenAI, Anthropic, Cohere, and Meta—to ensure their models do not have “hazardous capabilities,” such as creating biological or nuclear weapons or facilitating cyber attacks. Developers must also conduct safety testing and report their findings to a newly-created state body.
Andrew Ng, a prominent AI figure, criticized the bill, arguing it imposes enormous liabilities and stifles innovation. Elon Musk and former OpenAI staff have previously expressed concerns about AI’s potential risks to humanity.
The bill, co-sponsored by the Center for AI Safety (CAIS), aims to introduce basic safety evaluations for large AI models. While Senator Scott Wiener, who introduced the legislation, describes it as a “light-touch bill,” critics claim it is overly restrictive and could drive AI businesses out of California.
The tech sector argues that the bill will hinder open-source AI models, making developers liable for bad actors who misuse their models. In response, Wiener plans to amend the bill to exclude open-source models from the “kill switch” requirement and focus on models costing at least $100 million to train.
As AI regulation becomes a global issue, with efforts seen from the U.S. and U.K. governments, the debate over California’s AI bill continues to evolve.
