By 
Anthropic, a leading AI developer, has expressed support for California’s revised AI regulation bill, SB 1047, stating that the benefits of the legislation likely outweigh the associated costs. This marks a significant shift in sentiment towards the bill, which aims to regulate the development and deployment of artificial intelligence across the state.
San Francisco-based Anthropic, backed by industry giants Amazon and Alphabet, is a competitor to OpenAI, the creator of ChatGPT. The proposed bill, spearheaded by State Senator Scott Wiener, mandates rigorous safety testing for advanced AI models that either exceed $100 million in development costs or require substantial computing power. Additionally, AI developers in California would be required to incorporate a “kill switch” to deactivate AI models if they malfunction. The bill also empowers the state attorney general to take legal action against non-compliant developers.
Importance of the Bill
Senator Wiener recently amended the bill, incorporating feedback from tech companies, including Anthropic. One significant revision was the removal of a proposed government AI oversight committee, a move intended to address concerns from the tech industry.
Despite these changes, many tech companies remain cautious. Both Alphabet’s Google and Meta have voiced concerns, with Meta warning that the bill could create an unfavorable environment for AI innovation in California. OpenAI has argued that AI regulation should be handled at the federal level, suggesting that the state bill could lead to legal uncertainties.
Anthropic’s Perspective
Anthropic’s CEO, Dario Amodei, acknowledged the improvements in the revised bill in a letter to California Governor Gavin Newsom dated August 21. “In our assessment, the new SB 1047 is substantially improved, to the point where we believe its benefits likely outweigh its costs,” Amodei stated. He also noted that the company’s initial concerns about the bill potentially stifling innovation due to the fast-evolving nature of AI have been largely alleviated in the amended version.
However, Amodei did express that certain aspects of the bill remain “concerning or ambiguous,” indicating that while the revisions are a step in the right direction, some uncertainties still linger for AI developers.
As California moves forward with this landmark legislation, the responses from major AI stakeholders like Anthropic highlight the complex balance between fostering innovation and ensuring responsible AI development.
Comments