California’s AI Safety Bill Passes the Legislature, but Still Needs One More Thing

A controversial plan to control AI in California has been passed by the state legislature and is now on its way to Governor Gavin Newsom to be signed into law. Senate Bill 1047, which is also called the Safe and Secure Innovation for Frontier Artificial Intelligence Models Act, wants to make sure that advanced AI systems made in the state are tested for safety and are supervised.

The bill was written by Democratic Senator Scott Wiener of San Francisco. It requires companies that spend more than $100 million on training AI models or $10 million on changing them to do thorough safety tests. For example, these tests are meant to find possible dangers, such as the AI’s ability to cause security holes or help make dangerous weapons. The law also requires a “kill switch” to be built in so that AI systems can’t be controlled when they get out of hand, and it gives the state attorney general the power to sue coders who don’t follow the rules.

The bill has caused a lot of discussion, even though it passed the senate. Some former OpenAI workers and AI expert Yoshua Bengio are in favor of the bill because they say it is needed to keep AI from becoming a major threat to society. Their view is that the law will make sure that advances in AI are in line with public safety.

Other than that, big tech companies like Google, Meta, and OpenAI, along with other stars in the field, have strongly been against the plan. They say that the law could stop people from coming up with new ideas, especially in California, which is a center for AI development. Some people are also worried that the costs of following the rules could hurt smaller businesses and make it take longer for open-source AI tools to be released.

Politicians are also split over the bill. Some people, like Elon Musk, have spoken out in support of the bill, but others, like Nancy Pelosi, are worried that it could hurt California’s tech business. Now it’s up to Governor Newsom to decide whether to cancel the bill or sign it into law. His choice will have big effects on how AI is regulated in California in the future. While the debate goes on, other AI-related bills are being thought about in the state. These include steps to stop deepfakes in polls and stop computers from discriminating against people automatically. How these political attempts turn out will affect how AI is developed and regulated in California in the future.