In a landmark move, New York state lawmakers have passed the RAISE Act, a pioneering bill designed to mitigate the risks posed by advanced AI models from major players like OpenAI, Google, and Anthropic. This legislation aims to avert catastrophic scenarios, such as large-scale injuries, fatalities, or damages exceeding $1 billion.

The RAISE Act's passage is a significant moment for the AI safety movement, which has struggled to gain traction amidst Silicon Valley's rapid innovation and the previous administration's focus on speed over safety. Prominent advocates like Nobel laureate Geoffrey Hinton and AI research pioneer Yoshua Bengio have lent their support to the Act, which could set a precedent for legally mandated transparency in AI labs across the nation.

Senator Andrew Gounardes, a co-sponsor of the bill, has emphasized that the RAISE Act is crafted to avoid stifling innovation, a critique often leveled at California's SB 1047, which faced a veto. Gounardes warns, "The window to put in place guardrails is rapidly shrinking given how fast this technology is evolving," highlighting the urgency of the situation.

The RAISE Act now awaits the decision of New York Governor Kathy Hochul, who has the power to sign it into law, demand amendments, or veto it. If enacted, the bill would compel the world's largest AI labs to publish comprehensive safety and security reports and report any safety incidents. Companies failing to meet these standards could face civil penalties up to $30 million, as empowered by the Act.

The bill targets the world's largest companies, whether based in California or China, applying to those with AI models trained on over $100 million in computing resources, a threshold likely to cover all current AI models and those available to New York residents. While the RAISE Act shares some provisions with SB 1047, it addresses previous criticisms by not requiring a "kill switch" on AI models and not holding companies accountable for post-training harms.

Despite this, Silicon Valley has shown significant resistance to New York's AI safety bill, as noted by Assemblymember and co-sponsor Alex Bores. He argues that the RAISE Act will not hinder tech innovation, contrary to industry objections.

Critics, including Andreessen Horowitz general partner Anjney Midha, have called the RAISE Act "stupid," claiming it would hurt the US in the global AI race. Anthropic, an AI lab advocating for federal transparency standards, has yet to take an official stance on the bill, with co-founder Jack Clark expressing concerns over its broad implications for smaller companies.

Senator Gounardes反驳s these criticisms, asserting that the bill was designed to exclude small companies. Major tech companies like OpenAI, Google, and Meta have not yet responded to requests for comment.

Another concern is that AI developers might refrain from offering their most advanced models in New York. Assemblymember Bores dismisses this, stating that the RAISE Act's regulatory burden is light and should not force tech companies to withdraw their products from New York, considering the state's significant economic influence. He adds, "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."