On Thursday, New York state lawmakers passed the RAISE Act, a significant piece of legislation designed to mitigate the risks associated with frontier AI models developed by major tech companies like OpenAI, Google, and Anthropic.
The act specifically aims to prevent scenarios that could result in catastrophic outcomes, such as the death or injury of over 100 individuals or damages exceeding $1 billion.
The passage of the RAISE Act marks a notable victory for the AI safety movement, which has struggled to gain traction in recent years amid a prevailing focus on rapid innovation prioritized by Silicon Valley and the administration of President Donald Trump.
Proponents of the bill, including renowned figures in AI research such as Nobel laureate Geoffrey Hinton and pioneer Yoshua Bengio, have advocated for the RAISE Act as a critical step toward establishing regulatory standards in the fast-evolving field of artificial intelligence.
If enacted, this bill would introduce America’s first set of legally mandated transparency standards for frontier AI laboratories, a move that has generated considerable interest and some criticism.
New York State Senator Andrew Gounardes, one of the bill’s co-sponsors, emphasized that the RAISE Act was intentionally crafted to avoid stifling innovation among startups and academic researchers, a point of contention that had marred California’s similar but ultimately vetoed legislation, SB 1047.
Senator Gounardes stated, “The window to put in place guardrails is rapidly shrinking given how fast this technology is evolving. The people that know [AI] the best say that these risks are incredibly likely […] That’s alarming.”
The RAISE Act now awaits the decision of New York Governor Kathy Hochul, who can sign the bill into law, return it for amendments, or opt to veto it.
Should it become law, the RAISE Act would necessitate that the largest AI labs produce thorough safety and security reports regarding their frontier AI models.
Furthermore, it would mandate the reporting of safety incidents related to AI model behavior or any unauthorized access to AI models.
Failure to comply with these standards could lead to civil penalties for tech companies of up to $30 million, as authorized by the bill.
The RAISE Act is specifically targeted at the largest AI firms, irrespective of their geographic location.
This includes both California giants like OpenAI and Google, as well as international competitors such as DeepSeek and Alibaba.
Transparency requirements would apply exclusively to companies whose AI models have been trained using resources exceeding $100 million, making it relevant to the vast majority of current AI models available today.
Nathan Calvin from Encode, who played a role in drafting both the RAISE Act and the earlier SB 1047, indicated that this legislation was carefully designed to address some of the criticisms levelled against past efforts.
Notably, the RAISE Act does not mandate the inclusion of a ‘kill switch’ in AI models nor does it hold companies responsible for critical harms linked to the use of post-trained frontier AI models.
Amidst the bill’s progression, notable pushback from Silicon Valley and criticism from some legislators have surfaced.
For instance, Assemblymember Alex Bores, another co-sponsor of the RAISE Act, described the resistance as expected, while asserting that the legislation does not impose limitations that would hinder tech innovation.
Anjney Midha, a general partner at Andreessen Horowitz, criticized the RAISE Act on X, calling it “another stupid, stupid state level AI bill that will only hurt the US at a time when our adversaries are racing ahead.”
Earlier this month, Anthropic, an AI lab emphasizing safety, had called for federal transparency standards but has not yet taken an official position on the RAISE Act.
Co-founder Jack Clark expressed concern that the broad definitions in the bill could disproportionately impact smaller companies.
In response to this concern, Senator Gounardes maintained that the bill was designed with smaller companies in mind and should not impose undue burdens upon them.
Critics often worry that the RAISE Act could lead AI developers to withhold their most advanced technologies from the New York market, a concern also raised against SB 1047 and evident in Europe’s stringent tech regulations.
Bores, however, contended that the regulatory requirements of the RAISE Act are not excessively burdensome and should not deter companies from offering their products in New York.
With New York ranking as the third largest GDP in the U.S., any withdrawal strategy would likely be met with serious consideration by tech firms, according to Bores.
He concluded, “I don’t want to underestimate the political pettiness that might happen, but I am very confident that there is no economic reason for [AI companies] to not make their models available in New York.”
image source from：https://finance.yahoo.com/news/york-passes-bill-prevent-ai-220917739.html
