The Veto of SB 1047
Governor Gavin Newsom of California has vetoed Senate Bill 1047, a proposed legislation that aimed to establish the most stringent legal framework for artificial intelligence (AI) in the United States. The bill, which reached the governor’s desk in late August, required action by September 30th. Despite its intentions to regulate AI development and deployment, Newsom decided against signing it into law, citing concerns about its approach and potential unintended consequences.
Key Points of the Vetoed Bill
- SB 1047 would have applied to AI companies operating in California with models costing over $100 million to train or $10 million to fine-tune.
- The bill proposed requirements such as implementing a “kill switch” and establishing protocols for testing to mitigate risks like cyberattacks or pandemics.
- It included protections for whistleblowers and allowed the Attorney General to sue for damages caused by safety incidents.
- The legislation underwent changes during its development, including the removal of proposals for a new regulatory agency and the power for the state attorney general to sue developers preemptively.
Implications and Reactions
The veto of SB 1047 has significant implications for the AI industry and its regulation. Governor Newsom expressed concerns that the bill’s approach might not effectively address the real threats posed by AI technology. He argued that the legislation could provide a false sense of security and potentially hinder innovation in the field.
The decision has sparked diverse reactions from various stakeholders. Some tech companies, including OpenAI and members of the Chamber of Progress coalition, had opposed the bill, arguing it would impede innovation. Others, like Anthropic, had shown support for the amended version of the bill.
Senator Scott Wiener, the bill’s main author, expressed disappointment with the veto, viewing it as a setback for those advocating for oversight of large corporations developing critical technologies. The veto leaves a regulatory vacuum in the U.S., as there are currently no binding restrictions on companies developing powerful AI technologies at the federal level.
Sources: techcrunch.com, theverge.com
Image Source: techcrunch.com
