SB 53: California has a new AI law

Although Donald Trump has banned any regulation that could be detrimental to AI companies, there is now a new AI law in California. It is far from the only law regulating AI in the US, but it is new and specifically targets AI companies, most of which are based in San Francisco, California.

SB 53 is the second bill in California to regulate AI. However, the governor rejected the first bill. Now Gavin Newsom has signed version two after it was already passed by the State Legislature. This makes the law valid. Newsom said in a statement: “California has proven that regulation can be introduced to protect the community while the AI industry continues to grow.”

The law is primarily concerned with transparency requirements for AI providers. And the safety precautions that OpenAI, Meta, Anthropic, Google, and others take for their AI models. Recently, there have been numerous reports about alleged relationships between humans and chatbots and also that chatbots can harm children and young people. In response, OpenAI recently introduced new safety mechanisms and a parent mode.

In addition, SB 53 states that security incidents and incidents relating to criminal offenses must be reported by AI companies. This includes, for example, fraudulent behavior by an AI model itself. Something that is not even required by the European AI Act. There is also legal protection for whistleblowers from AI companies.

AI providers lobby against AI regulation

The reactions to the law from AI providers are, of course, rather negative. It has long been criticized that there are too many different laws regulating AI in the individual states of the USA, some of which contradict each other. According to AI providers, the Californian AI law certainly threatens to prevent innovation. This in turn would be particularly damaging to the US economy in the race with China, according to lobbying circles. Meta, for example, recently founded an interest group specifically for California, which was apparently intended to prevent Newsom's signature. There are also such PACs (Political Action Committees) in other states.

In California, there is even another piece of legislation that is supported by Democrats and Republicans. This deals with chatbots that explicitly have a friend mode. Safety mechanisms are also required here, and the law also stipulates that AI providers can be held liable if the mechanisms do not work.

In New York, a similar bill to SB 53 is waiting for the governor there, Kathy Hochul. In Texas, Attorney General Ken Paxton is particularly opposed to AI companies. He is concerned about the safety of children and young people.

(emw)