California Takes Major Step Towards AI Safety Verification with New Bill

California Moves Forward with Independent AI Safety Verification

In a monumental decision, the California Legislature has passed SB 813, aimed at revolutionizing the way artificial intelligence (AI) safety is assessed. This bill, sponsored by Fathom and authored by Senator Jerry McNerney, sets the groundwork for the establishment of Independent Verification Organizations (IVOs) to scrutinize AI safety claims and enhance public confidence in AI technologies.

Fathom, a nonprofit organization dedicated to creating a governance structure for AI, has been at the forefront of advocating for the IVO model encapsulated in SB 813. These organizations would comprise qualified external experts capable of evaluating the risks associated with various AI systems through credible testing and assessments, fostering a more accountable environment.

If this bill receives the signature of Governor Newsom, it will mark a pivotal first step by any state to build an ecosystem focused on AI evaluation. With rising concerns about the implications of AI technology—from ethical dilemmas to potential risks for children and families—this legislation is designed to ensure independent oversight rather than relying solely on self-reports from tech companies.

California previously spearheaded AI transparency with SB 53, which mandated large AI developers to disclose their safety frameworks and report critical incidents. However, SB 813 is a significant leap forward as it introduces a system where independent assessments can validate the safety of AI technologies.

Andrew Freedman, CEO of Fathom, expressed enthusiasm about this legislative development, stating, "This independent verification model empowers qualified experts to continuously evaluate AI capabilities, ensuring that assessment practices evolve alongside technological advancements. This initiative offers not only a pathway for public trust in AI but also extends an opportunity for businesses to demonstrate their commitment to safety."

The call for accountability in AI has never been more pressing as the technology learns to integrate deeper into everyday life. Americans—be they in educational settings, enterprises, or households—are eager for clarity on how to harness the benefits of AI responsibly and safely. The IVO model proposed by SB 813 aims to build a robust framework that instills confidence in AI's role in society, paving the way for similar systems across other states.

As discussions in Congress and legislative bodies in various states regarding the necessity for third-party evaluations of AI technology continue, California stands out as a pioneering state taking concrete steps to institutionalize this need. Fathom's active role in promoting the independent verification model positions the organization as a trailblazer in the effort to safeguard public interests in the realm of AI development.

In addition to legislative progress, Fathom continues to collaborate with policymakers, technical experts, and industry leaders. Their objective is to fortify California's stance on AI governance while creating a replicable model for other states to follow, ensuring that safety and accountability remain at the forefront of AI innovation.

As we advance into an era increasingly shaped by artificial intelligence, it is essential that we develop systems that not only leverage the technology's capabilities but also prioritize the safety and well-being of the public. California's SB 813 stands as a testament to the shared commitment to fostering ethical AI practices and establishing a precedent for nationwide standards.

For more information about Fathom and their ongoing initiatives, visit fathom.org.