California Bill Imposes New Rules On AI

The California State Assembly has approved the Safe and Secure Innovation for Frontier Artificial Intelligence Models Act (SB 1047) after many amendments were made earlier this month.

Introduced by State Senator Scott Wiener, SB 1047 aims to mitigate risks associated with advanced AI, such as autonomous weapons or large-scale cyberattacks. Furthermore, as reported by Business Insider (BI), the bill requires companies developing AI models to implement safety measures, including kill switches and third-party testing.

Key provisions of SB 1047 include strict safety protocols for developers of “covered models,” which are large-scale AI models that require significant computing power to train, and have the potential to cause significant harm.

These protocols require developers to implement comprehensive security measures to address risks like unintended consequences or malicious use. In addition, developers must undergo annual independent audits to ensure compliance and report any AI safety incidents to the Attorney General within 72 hours.

The bill also places responsibilities on computing cluster operators, requiring them to identify customers training these high-risk models and take steps to mitigate potential dangers.

SB 1047 has sparked significant debate within the tech industry. While proponents argue that it is essential to safeguard the public from advanced AI’s potential dangers, opponents raise concerns about its potential impact on innovation.

Critics also argue that the regulations could burden companies, limiting their ability to compete with larger tech giants. However, Senator Wiener defended the legislation, emphasizing that it targets companies spending over $100 million on AI development, exempting smaller startups, as reported by Fortune.

Elon Musk, X CEO and founder of xAI, expressed his support for the bill. He acknowledged its potential to upset some but emphasized the need for regulation to prevent public risks.

However, other major AI companies and experts have voiced opposition. For example, the LA Times reports that OpenAI’s Chief Strategy Officer Jason Kwon wrote in a letter to Wiener, “A federally-driven set of AI policies, rather than a patchwork of state laws, will foster innovation and position the U.S. to lead the development of global standards.”

Additionally, Stanford professor Fei-Fei Li, an AI expert, argued that while the legislation is well-meaning, it could have significant unintended consequences, including damaging the open-source community and limiting academic access to essential AI models.

Despite the controversy, SB 1047 also proposes creating CalCompute, a public cloud computing cluster, to support responsible AI development and provide researchers and developers with access to necessary resources.

While the bill represents a significant step toward regulating AI in California, critics and supporters alike acknowledge that the rapidly evolving AI landscape may require further revisions to the law to address emerging challenges and opportunities.