California passes first AI safety law in the US

California has just become the first US state to enact artificial intelligence (AI) safety legislation, marking a turning point in efforts to control rapidly developing AI technology.

Governor of California (USA) Gavin Newsom - Photo: AFP

According to Politico , on September 29, California Governor Gavin Newsom signed into law Bill SB 53, requiring large AI companies such as OpenAI and Meta to publicly disclose safety procedures, report serious incidents, and protect whistleblowers.

Under the new law, businesses must publicly submit security protocols—in redacted form to keep intellectual property confidential—and report incidents within 15 days if there is a major risk such as cyberattacks, AI-powered weapons, or loss of model control.

In particular, the law requires companies to report dangerous fraudulent behavior by AI systems during the testing phase – a provision considered a “ world first”.

SB 53 also lays the groundwork for a state-run cloud computing system called CalCompute, and provides whistleblower protections.

According to AFP, California's approach differs from the European Union's AI Act, which only requires businesses to report privately to the government . SB 53, on the other hand, requires public disclosure to increase accountability.

The new bill is seen as a major victory for state Senator Scott Wiener, the bill's author, after the previous version (SB 1047) was vetoed last year due to strong opposition from the tech industry.

This time, Mr. Wiener narrowed the bill, working with the Governor's office and gaining support from a number of companies like Anthropic (developer of the chatbot Claude), while OpenAI and Meta did not publicly oppose it.

“With a technology as transformative as AI, we have a responsibility to support innovation but also to set reasonable guardrails to understand and mitigate risks. With this legislation, California is once again a world leader in both technological innovation and safety,” Wiener said.

The move comes as the Trump administration pushes for AI development to compete with China, while lawmakers in the US Congress and other states like New York are considering similar bills.

However, lobbying group Chamber of Progress and many investors have spoken out against the bill, fearing that each state will have its own standards, which will cause “legal chaos,” and argue that the bill will stifle California’s innovation economy and put start-ups at a disadvantage compared to large corporations.

PUBLIC REVELATION

Source: https://tuoitre.vn/california-thong-qua-luat-an-toan-ai-dau-tien-tai-my-20250930105812564.htm