Veto against California AI law – Silicon Valley prevails

It was created with good intentions, but does not achieve the goal – this is what California’s responsible governor Gavin Newsom writes about the planned AI law SB-1047. The short name is Senate Bill 1047, the long title is “Safe and Secure Innovation for Frontiers Artificial Intelligence Models Act”. Newsom used his veto power. AI experts and companies had already expressed sharp criticism of the legislation — albeit of a different nature than Newsom’s.

Advertisement

In a public statement, the Governor explains his reasons. The law does not take into account the environment in which AI is used – whether in areas of high risk to society or in entirely mundane tasks. Newsom even fears that SB-1047 could give people a false sense of security. Because only large AI models are regulated. He warns that smaller, specialized models can be at least as dangerous.

Lion vs. Stock Photographer: Court allows records

Governor still demands regulation

Still, he believes AI needs to be regulated, Newsom writes. There should be protocols in place for monitoring, security, and consequences for misuse of AI. A law should also be adaptable because AI is a technology that is changing and evolving. “To be clear – I agree with the author – we cannot wait for a major disaster before taking action to protect the public.”

SB-1047 also proposed creating a state board to oversee AI models and providers. Researchers and companies criticized, among other things, the requirement of a kill switch for models of a certain size. AI expert Fei-Fei Li says this will especially prevent the open source community from growing. California is home to 32 of the world’s top 50 AI companies, Newsom writes. Newsom also believes it is right for US states to move forward with regulation while Congress takes longer.

The European AI regulation, unlike the Californian proposal, regulates AI using a risk-based approach. Many AI applications can be easily used, such as spam filters for email. Other applications, such as social scoring, are prohibited. Still others must meet certain transparency and documentation obligations.

(EMW)