- Anthropic said Thursday that the benefits of California’s revised bill, which seeks to regulate how artificial intelligence is developed and deployed in the state, could outweigh the costs.
- San Francisco-based Anthropic is a competitor to ChatGPT maker OpenAI and is backed by Amazon and Alphabet.
OUR TAKE
Senator Wiener recently revised SB 1047 to appease tech companies, which relied in part on Anthropic’s input. The revised bill removes the provision for a governmental AI oversight board, which was initially feared to potentially stymie innovation due to the rapid growth of the field, but this fear has been greatly reduced in the revised version.
— Iydia Ding, BTW reporter
What happened
California’s proposed AI regulatory bill, SB 1047, introduced by state Senator Scott Wiener, a Democrat, would require safety testing of many state-of-the-art AI models that cost more than $100 million to develop or require a certain amount of computational power, Anthropic said Thursday. California’s revised bill seeks to regulate the way AI in the state isthe way it is developed and deployed, and that the benefits could outweigh the costs. San Francisco-based Anthropic is a competitor to ChatGPT maker OpenAI and is backed by Amazon and Alphabet.
AI software developers operating in the state would need to outline ways to shut down AI models if they go wrong, effectively a termination switch. The bill would also give the state’s attorney general the power to sue developers if they don’t comply. “Based on our assessment, the new SB 1047 is substantially improved, so much so that we believe its benefits may outweigh its costs. However, we’re not sure about this, and some aspects of the bill seem worrisome or ambiguous to us.” Anthropic CEO Dario Amodei said in an August 21 letter to California Governor Gavin Newsom.
Also read: Authors sue Anthropic for copyright infringement over AI training
Also read: Claude AI chatbot creator Anthropic sued for copyright infringement
Why it’s important
Senator Wiener recently revised SB 1047 to appease tech companies, which relied in part on Anthropic’s input. The revised bill removes the provision for a governmental AI oversight board, which was initially feared to potentially stymie innovation due to the rapid growth of the field, but this fear has been greatly reduced in the revised version.
Tech companies developing artificial intelligence-which can respond to prompts with fully formatted text, images, or audio and run repetitive tasks with minimal intervention-are largely opposed to the bill.
Alphabet’s Google and Meta expressed concerns in a letter to Wiener, with Meta stating that the bill could potentially make the state hostile to AI development and deployment. OpenAI has said that AI should be regulated by the federal government, and that the California bill creates an uncertain legal environment. But as AI technology advances at a rapid pace, the laws and regulations will be refined and gradually adapted to the realities of the situation as they are implemented.
