Innovative Steps Forward: New York's RAISE Act Seeks to Rein In frontier AI, Igniting an Era of Heightened Transparency

In a big win for proponents of AI safety, New York state lawmakers have successfully passed a key bill that seeks to create a fail-safe approach to frontier AI models and their potential hazards. Designed to address concerns of catastrophic outcomes, such as those causing death, injury, or over $1 billion in damages, the proposed law could represent a radical shift in the sector. Standing as a potentially much-needed check on giants like OpenAI, Google, and Anthropic, the RAISE Act is pleading for prudence alongside progress in AI development. The move signals a growing urgency to control and monitor cutting-edge AI, even as Silicon Valley and previous administrations have prioritized rapid innovation over entrenched safety measures. The RAISE Act has already garnered support from leading figures in the tech community, including Nobel laureate Geoffrey Hinton and AI research pioneer Yoshua Bengio. If the bill becomes law, it would set a groundbreaking precedent by establishing the first legally mandated transparency standards for frontier AI labs in the United States. Its recent success indicates a hard-won victory for the AI safety movement, which has repeatedly found itself sidelined in favor of speed and innovation. Regardless of past losses, advocates are now having their most significant moment yet. Unquestionably, this legislation seems to aim at striking a balance between safety guarantees and innovation incentives. New York state Senator and bill co-sponsor Andrew Gounardes has explicitly stated his intent to design the RAISE Act in a way that doesn’t deter innovation among startups or academic researchers – a primary critique of similar past bills. Companies that neglect to meet these standards may face civil penalties up to $30 million, courtesy of New York’s attorney general due to the power vested by the RAISE Act. While global tech giants, whether from California or China, are under the purview of this new regulation, the transparency requirements mainly apply to companies that have trained their AI models using over $100 million in computing resources. It excludes models that are not being offered to New York residents. Despite its laudable goals and adjustments, the RAISE Act has, predictably, faced substantial resistance from Silicon Valley. However, New York state Assemblymember Alex Bores, who is also among the co-sponsors of the RAISE Act, argues that the regulation would not hamper the innovation potential of tech companies. As the bill awaits Governor Kathy Hochul’s signature to become law, the world is watching. Irrespective of the final outcome, the RAISE Act underlines an urgent need to shift gears from unrestrained innovation to responsible, transparent progress in AI development. The bill now rests on the brink of history, with the potential to reshape how we approach the development and deployment of frontier AI models.