New York passes a bill to prevent AI-fueled disasters

Legislers in the state of New York an account approved On Thursday this wants to prevent Frontier AI models from OpenAi, Google and Anthropic to contribute to disaster scenarios, including the death or injury of more than 100 people, or more than $ 1 billion compensation.

The approval of the Raise Act represents a victory for the AI ​​safety movement, which has lost ground as Silicon Valley and the Trump administration in recent years have given priority to speed and innovation. Proponents of safety, including Nobel Prize winner Geoffrey Hinton and AI research spy Yoshua Bengio have defended the Raise Act. If it becomes the law, the bill would determine the first series of legally required transparency standards for Frontier AI Labs.

The Raise Act has some of the same provisions and goals as the controversial AI safety account of California, SB 1047, which eventually became veto. However, the co-sponsor of the bill, Senator Andrew Gounardes in New York, told TechCrunch in an interview that he deliberately designed the Rais Act, so that it does not cool up under startups or academic investigators, a common criticism of SB 1047.

“The window to set up guardrails is shrinking quickly, given how quickly this technology evolves,” said Senator Gouden. “The people who know [AI] say best that these risks are incredibly likely […] That is alarming. “

The Raise Act is now on the way to the Bureau of Governor of New York Kathy Hochul, where it could sign the law in the law, can send it back for changes or pronounced a veto.

If signed by law, the AI ​​safety account of New York would require that the world’s largest AI Laboratories publish thorough safety and security reports on their frontier AI models. The account also requires AI Laboratories to report safety incidents, such as about AI model behavior or bad actors that steal an AI model, they were allowed to take place. If technology companies do not live up to these standards, the Raise Act enables the attorney general of New York to bring civil fines of a maximum of $ 30 million.

The Raise Act aims to regulate the largest companies in the world – whether they are located in California (such as OpenAi and Google) or China (such as Deepseek and Alibaba). The transparency requirements of the bill apply to companies whose AI models have been trained with more than $ 100 million in computer sources (apparently, more than any AI model available today), and are made available to residents of New York.

Although comparable to SB 1047 in some ways, the raise act was designed to tackle criticism of earlier AI security accounts, according to Nathan Calvin, the vice-president of the state affairs and general adviser at ENCODE that worked on this account and SB 1047. Remarkable, the Razenwet’s Nietly Devel Devel Devel Devel Devel Devel Devel Devel Devel Devel Devel Devel Devel Devel Devel Developed Use companies in De Roos for critical complaints.

Nevertheless, Silicon Valley has considerably pushed back on the AI ​​Safety Bill in New York, the New York State Assembly member and co-sponsor of the Raise Act Alex Bores told WAN. Bores did not mention the industrial resistance surprising, but claimed that the Raise Act would in no way limit the innovation of technology companies.

“The NY Raise Act is another stupid, stupid state level AI Bill who will only harm the US at a time when our opponents race ahead” Friday post on X. Andreessen Horowitz and Startup Incubator Y Combinator were some of the brightest opponents of SB 1047.

Anthropic, the safety-oriented AI laboratory that called for federal transparency standingNDards for AI companies Earlier this month, no official position in the account has achieved, said co-founder Jack Clark in a Friday post on X. However, Clark has expressed a few complaints about how broad the Razen Act is and noted that it can pose a risk for ‘smaller companies’.

When he was asked about Anthropic’s criticism, Senator Goundes told WAN that he thought it “missed the goal”, and noted that he designed the bill not to apply for small companies.

OpenAi, Google and Meta did not respond to WAN’s request for comment.

Another common criticism of the Raise Act is that AI model developers would simply not offer their most advanced AI models in the state of New York. That was a similar criticism of SB 1047, and it is largely what is being played in Europe thanks to the heavy regulations of the continent for technology.

Assembler -Bores told WAN that the legal burden of the Rais Act is relatively light and should therefore not require technology companies to stop managing their products in New York. Given that New York has the third largest GDP in the US, the retirement of the state is not something that most companies would take lightly.

“I don’t want to underestimate the political narrowness that could happen, but I am confident that there is no economic reason [AI companies] To make their models not available in New York, “said Assembly member Bores.