New York's RAISE Act: A Pioneer in AI Model Regulation

In a pioneering move, Assembly Member Alex Bores from New York, a Democrat with a background in computer science, is intent on reviving the principles of the failed California AI safety bill, SB 1047. By proposing the RAISE Act (Responsible AI Safety and Education), Bores aims to craft a legislative framework to regulate the most advanced AI models, mirroring some of the intentions of its California predecessor.

Key Features of the RAISE Act

The RAISE Act is currently in draft form and seeks to address several concerns that impeded the success of SB 1047. A central component of the New York bill is the requirement for AI companies to develop comprehensive safety plans that address both development and deployment of AI models. These plans should include cybersecurity measures, assessment of risks associated with models both before training and after deployment, and protocols for preventing unauthorized access.

A noteworthy protection mechanism within the act is for whistleblowers: individuals who report potential critical harm from AI models—including uses such as creating weapons or causing large-scale harm—are safeguarded from retaliation.

Furthermore, unlike SB 1047, the RAISE Act does not propose new institutions such as the Board of Frontier Models, nor does it suggest the creation of a public cloud computing cluster. It also evades contentious requirements such as a model “kill switch,” simplifying its approach to focus specifically on critical harms rather than broader AI concerns like bias or job displacement.

The Challenge of Regulation

The proposal of the RAISE Act highlights ongoing debates in AI regulation. Critics in the AI sphere, like those from the AI Now Institute, warn that focusing solely on catastrophic risks might overlook everyday risks posed by AI, such as bias and environmental impacts. Yet Bores stands firm, recognizing that the bill’s current focus is on frontier models that represent the future edges of AI capability.

The introduction of the RAISE Act illustrates the significant challenges in creating AI regulations—particularly as major industry players like Google and Meta previously opposed SB 1047.

Conclusion: A Step Toward Responsible AI Governance

As states like New York take steps to lead in AI regulation, the conversation initiated by bills such as SB 1047 and now the RAISE Act emphasizes the need for these powerful technologies to be governed responsibly. Despite the bill still being in its infancy, it is a significant stride towards crafting robust frameworks that anticipate AI’s future challenges while learning from past legislative efforts.

Ultimately, the RAISE Act is a harbinger of ongoing discussions on how best to regulate frontier AI models in a rapidly advancing technological landscape. While it awaits formal introduction and subsequent debates, its foundations could inspire other states and possibly national-level actions on AI regulation, highlighting New York’s key role in this tech governance evolution.