California one step away from putting AI on a short leash

AI tech companies with operations in California might have to relocate or rethink their approach to research in the field soon if Governor Gavin Newsom decides to approve the AI safety bill, SB 1047. According to it, testing advanced AI models has to be done after the developers implement a kill switch. While some argue that this is a welcome safety measure, the critics see it as a severe limitation that would stop innovation and ultimately drive many companies out of the state.

Elon Musk, CEO of both Tesla and AI firm xAI, is in favor of the bill. On the other hand, huge tech names such as Google, Meta, and OpenAI are concerned that the limitations that come with this bill are not worth the added security against a possible yet quite unlikely, at least for now, AI threat. If the bill passes, the state attorney general would be able to sue non-compliant developers. Third-party audits and whistle-blower protection are also mentioned in the legislation that needs a final green light from California's Governor Gavin Newsom.

The decision resting on Newsom's shoulders is not an easy one, but he has until September 30th to decide. On the bright side, even if SB 1047 gets through, it should be highlighted that safety testing is required for AI models with costs of development over $100 million or those requiring a certain amount of computing power, so most AI startups would not need to bother with the kill switch implementation.

Those interested in AI can grab BK's Artificial Intelligence and the Human Mind: A Radical New Old Science of the Human Mindoff Amazon for $9.99 in Kindle format, $32.21 in hardcover shape, or $17.95 (paperback).

The above is the detailed content of California one step away from putting AI on a short leash. For more information, please follow other related articles on the PHP Chinese website!