Legislators across the United States are working on implementing regulations to control the development and deployment of high-level artificial intelligence models.
AI developer Anthropic has joined the tug-of-war surrounding California senator Scott Wiener’s proposed bill, saying its “benefits likely outweigh its costs.”
SB 1047 developments
In a letter on Aug. 21 to California governor Gavin Newson, the CEO of Anthropic, Dario Amodei said that the legislation in question, Senate Bill (SB) 1047, has been “substantially improved,” though there are still some aspects which seem “concerning or ambiguous.”
“Our initial concerns about the bill potentially hindering innovation due to the rapidly evolving nature of the field have been greatly reduced in the amended version.”
SB 1047 was introduced by Wiener back in February 2024 and seeks to mandate and standardize safety testing for the most advanced AI models, which either cost over $100 million to develop or require a set amount of computing power.
According to Wiener’s bill, AI software developers within the state of California would need to provide outlines for the methods in which the models could be turned off in case of emergency. This is often referred to as a “kill switch.”
Related: OpenAI offers GPT-4o fine-tuning for companies in AI service shift
Developers of AI software operating in the state would need to outline methods for turning off the AI models if they go awry, effectively a kill switch. Additionally, the bill would grant the state attorney general the ability to bring legal action against the developers if they are found to be noncompliant with the law.
Tech companies chime in
The bill was recently revised by Wiener, which included influence from Anthropic, among other tech companies, in order to appease some of the complaints from Big Tech. One of the revisions included the nixing of the creation of a government-led AI oversight committee.
However, Wiener has also been vocal about industry opposition to his bill. On Aug. 22, the senator responded to a letter penned by OpenAI to Governor Newsom that opposed SB 1047.
Wiener called OpenAI’s concerns “unfounded,” “tired,” and “baseless.” He said the bill is essential for safeguarding public safety and national security and is perplexed by OpenAI’s current position due to its previous commitment to conducting such safety evaluations.
Along with OpenAI, Alphabet’s Google and Meta have also sent letters of concern to Wiener. Meta claims that the bill threatens to make California less hospitable for AI development and deployment.
Magazine: AI may already use more power than Bitcoin — and it threatens Bitcoin mining