California’s AI Safety Bill Sparks Debate Among Tech Leaders
California is on thecusp of becoming the first state to implement significant regulations on the development of powerful artificialintelligence (AI) systems. The controversial AI safety bill, SB 1047, passed the state Assembly on August 28th, sending itback to the Senate for a final vote before it reaches Governor Gavin Newsom’s desk.
The bill, introduced by California State Senator Scott Wiener, mandatesthat companies developing advanced AI models take reasonable precautions to ensure their technology doesn’t cause serious harm, defined as mass casualties or property damage exceeding $500 million. It requires companies to implement safety measures like kill switches to disable their AI systems, submit models for third-party testing, and provide whistleblower protection for employees who wish to report safety concerns. Failure to comply could result in legal action by the California Attorney General.
Wiener, a vocaladvocate for responsible AI development, argues that the bill is a simple, common-sense measure that codifies safety commitments already made by many large companies. He emphasizes the need for proactive measures to protect the public interest as AI technology rapidly advances. Innovation and safety can go hand-in-hand, and Californiais leading the way, he said in a statement.
However, SB 1047 has faced fierce opposition from numerous tech leaders, startups, and venture capitalists who view it as excessive government intervention in a nascent technology. They fear it could stifle California’s technological innovation and hinder its position as a global AIhub.
OpenAI, the leading AI research company, publicly opposed the bill last week, arguing that such policies should be implemented at the federal level rather than on a state-by-state basis. Other prominent critics include Congressman Ro Khanna and San Francisco Mayor London Breed, who share the tech industry’sconcerns about the bill’s potential to impede California’s AI leadership.
Despite the initial backlash, the bill gained support from some key figures in the AI field in the final stages of the voting process. Elon Musk, a vocal advocate for AI safety, surprisingly expressed his support for the bill on August 27th, though he acknowledged it was a tough decision that will make some people unhappy.
Anthropic, a safety-focused competitor to OpenAI, also cautiously endorsed the bill, stating that the benefits may outweigh the costs and that the provisions are practically implementable. Their support grew stronger after Wiener incorporatedseveral of their suggested amendments.
Wiener has defended the bill against critics, highlighting that its provisions only apply to companies spending over $100 million to train large models or over $10 million to fine-tune them, effectively exempting most smaller startups. He also emphasizes his support for federal legislation but acknowledges Congress’s historical slowness in regulating technology. He believes California has a responsibility to take the lead in the absence of national action.
The debate surrounding SB 1047 is far from over. OpenAI, the tech incubator Y Combinator, and the venture capital firm Andreessen Horowitz, all critics of thebill, have registered lobbyists to influence its fate. As the bill moves towards Governor Newsom’s desk, the future of AI regulation in California, and potentially the nation, hangs in the balance.
【source】https://www.ithome.com/0/792/094.htm
Views: 1