Controversy over California Bill SB 1047 aiming to regulate large frontier AI models in US

The introduction of a new bill in California, known as SB 1047, aimed at regulating large frontier AI models has ignited a heated debate within the tech industry. The bill, proposed by California state Senator Scott Weiner, has received opposition from various stakeholders including startup founders, investors, AI researchers, and advocates for open source software.

Weiner states that the bill aims to establish common-sense safety standards for developers of powerful AI systems. However, opponents of the legislation argue that it would stifle innovation and potentially damage the entire AI industry.

The bill, currently making its way through various committees after being passed by the California legislature, is set to have a final vote in August. If approved, it would mark the first major law regulating AI in a state that houses many tech giants.

Under the ‘Safe and Secure Innovation for Frontier Artificial Intelligence Systems Act’, large AI companies such as Meta, OpenAI, Anthropic, and Mistral would be held accountable for potential risks associated with their advanced AI models. The bill defines ‘large’ AI systems as those trained using significant computing power and resources.

One of the key provisions of the bill includes holding developers responsible for “critical harms,” such as the creation of weapons of mass destruction and cyberattacks on critical infrastructure. It also mandates the inclusion of a kill switch in AI models for emergencies.

While proponents of the bill, including figures like Geoffrey Hinton and Yoshua Bengio, support the regulation of AI to mitigate potential risks, critics argue that the legislation could hinder open source AI models and place excessive regulatory burdens on companies.

As the debate over the bill continues, both supporters and opponents are closely monitoring its progress and potential implications on the future of AI development in California.