New York launches statewide network to test AI safety
Governor Hochul established a new research center and a network of 80+ experts to independently audit frontier AI models.
The short version
- The New York Center for AI Responsibility and Research (NYCAIRR) will lead independent testing of frontier models.
- A network of more than 80 researchers will work across various institutions to evaluate AI security and behavior.
- Large AI developers must register with the state and report safety incidents starting in January 2027.
- The project is funded by a $30 million philanthropic gift, $25 million in state research capital, and operating funds.
Why it matters
This creates a formal mechanism for the government to verify AI safety claims rather than relying solely on company self-reporting. It sets a precedent for how states can use academic research to inform tech regulation.
A new hub for independent testing
Governor Kathy Hochul officially launched a statewide effort to monitor and test artificial intelligence safety on November 10, 2026. This initiative places the New York Center for AI Responsibility and Research (NYCAIRR) at Binghamton University in charge of the project. The center will focus on independent evaluations of frontier models, which are the most advanced and powerful AI systems currently in development.
NYCAIRR is designed to provide public officials with objective evidence to help them make informed decisions regarding AI accountability. The center will not only assess existing systems but also work to create new measurement tools and standards for future models. This research aims to support the responsible use of AI in high-stakes fields such as medicine, scientific discovery, and critical infrastructure.
The center will utilize advanced computing resources provided by Empire AI, a $500 million public-private initiative. This partnership is intended to support AI research that serves the public good. By establishing a dedicated research body, the state aims to create a scientific foundation for how AI is integrated into public life.
Building a statewide research network
To execute its mission, NYCAIRR will manage a network of more than 80 researchers who will remain at their home institutions. The network will be guided by interim co-directors Greg Morrisett of Cornell Tech and Jeannette Wing of Columbia University. They are tasked with overseeing the scientific agenda and building the infrastructure for statewide independent evaluations.
The state plans to recruit specialists in a wide range of technical fields to join this network. These include experts in mathematical theory, hardware security, privacy, and biosecurity. The goal is to understand how advanced systems operate and whether they behave as intended by their creators. The network will also look at how AI can be used responsibly in scientific research.
Governor Hochul emphasized that the state needs a way to verify the claims made by technology companies. She stated, "The stakes are too high to simply take AI companies at their word." By employing a diverse group of specialists, the state hopes to provide a rigorous check on the safety and security of frontier models.
Funding and organizational structure
The launch of NYCAIRR is supported by a combination of philanthropic, state, and private funding. A $30 million philanthropic gift from a donor group led by Tom Secunda provided the initial spark for the center. Additionally, the State University of New York provided a $25 million research capital investment to support the project's goals.
New York State is also providing the necessary operating capital to keep the center running. This financial backing allows the center to pursue foundational scientific research and train new AI safety experts. The November launch specifically adds the statewide network and independent-evaluation program to the original mission announced in early 2026.
The structure of the initiative is designed to separate research from regulation. While NYCAIRR provides the technical expertise and evidence, other state offices will handle the actual administration of the law. This distinction ensures that the scientific testing remains independent from the government's enforcement actions.
New mandates for AI developers
In tandem with the research launch, New York opened the RAISE Portal to facilitate the Responsible AI Safety and Education Act. This portal serves as a registration point for large frontier AI developers operating within the state. All developers must register with the state by January 1, 2027, to comply with new transparency and safety requirements.
The Department of Financial Services is housing a specific branch called DIGIT to oversee these rules. Large developers will face stricter obligations, including the requirement to publish safety frameworks beginning in 2027. All frontier developers will be required to report critical safety incidents and publish transparency reports.
Specific rules for large developers include reporting critical safety incidents to DIGIT within 72 hours. They must also file quarterly assessments regarding catastrophic risks and submit disclosure statements at least every two years. NYCAIRR will coordinate with DIGIT to review these published frameworks and suggest ways to strengthen the state's requirements.
What to watch
- How the research network's findings will influence specific state regulations.
- Whether large AI developers comply with the 72-hour incident reporting window.
Try it
Check the New York Department of Financial Services website for updates on the RAISE Portal and upcoming registration deadlines for AI developers.
Sources and links
- Governmentlaunched a statewide AI science and safety effort governor.ny.gov
- GovernmentSeptember 21, 2026 implementation announcement governor.ny.gov
- ReportingUnite.AI: the article we sourced
First reported by Unite.AI. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.