Scott Wiener on his fight to make Big Tech disclose AI’s dangers

California State Senator Scott Wiener continues his push for AI safety regulation with the introduction of a new bill, SB 53, following the controversial SB 1047, which faced significant backlash from Silicon Valley. The previous bill aimed to hold tech companies accountable for potential AI-related harms but was ultimately vetoed by Governor Gavin Newsom amid concerns that it could hinder innovation in the AI sector. As SB 53 awaits the governor’s decision, it appears to be gaining traction, with notable backing from companies like Anthropic and support from Meta, which believes the bill strikes a balance between necessary oversight and innovation. This shift in sentiment reflects a growing recognition within the industry of the need for safety protocols without stifling technological advancements. If enacted, SB 53 would mandate safety reporting requirements for major AI firms, specifically those earning over $500 million in revenue. These companies, including OpenAI, Anthropic, and Google, currently have no obligation to disclose their safety testing methods. While some AI labs voluntarily release safety reports detailing the potential misuse of their technology, the bill aims to impose more consistent and formalized expectations. The focus of SB 53 is on the most severe risks associated with AI, including the potential for causing human fatalities, cyberattacks, and the development of bioweapons. Unlike the previous bill, which included liability provisions, SB 53 emphasizes transparency and self-reporting, thus making it less contentious among tech leaders. Wiener has expressed skepticism regarding federal oversight, arguing that states must take the lead on AI regulation, especially given the current federal administration's shift towards prioritizing growth over safety. He believes California has a critical role in shaping responsible AI practices that protect public safety without curtailing innovation. The senator’s commitment to this cause is underscored by his acknowledgment of the complexities involved in regulating a rapidly evolving industry. In conversations with industry leaders, Wiener emphasizes the necessity of addressing catastrophic risks posed by advanced AI technologies. He acknowledges the diverse risks associated with AI, but asserts that SB 53 is strategically focused on those that could result in the most profound harm. As the debate surrounding AI safety continues, the outcome of SB 53 could mark a pivotal moment for regulation in the tech industry, setting a precedent for how states approach the challenges posed by AI advancements.