New York stands up its own AI testing center, with 80+ researchers

New York Gov. Kathy Hochul has launched the New York Center for AI Responsibility and Research (NYCAIRR), a state effort to independently test frontier AI systems and train a new generation of AI safety experts. Politico was first to report it, and the governor’s office says the center will bring together more than 80 scientists across the state.

A Binghamton hub run by Cornell Tech and Columbia leaders

According to the governor’s release, as published by LongIsland.com, the center is based at SUNY Binghamton University. Greg Morrisett, dean and vice provost of Cornell Tech, and Jeannette Wing, Columbia University’s executive vice president for research, are the interim co-directors. They will set the scientific agenda, oversee the independent evaluation work and build a statewide network of researchers who remain at their home institutions.

The stated job list is broad: test frontier systems for safety and security, run foundational research, and train safety experts. The center will also develop the tools, measurements and standards needed to assess more powerful systems later, and give public officials evidence for AI safety and accountability policy. At the State’s request it will supply assessments and evidence on frontier developers, and coordinate with the state’s DIGIT office (Digital Innovation, Governance, Integrity, and Trust) to review the safety frameworks developers publish and recommend stronger requirements.

A national recruiting drive is planned for six areas: foundational limits and mathematical theory, formal methods and verification, agentic and complex systems, hardware security and privacy, interpretability and alignment, and scientific AI and biosecurity.

Money and computing come from four places

The release lists a $30 million philanthropic gift from donors led by Bloomberg LP co-founder Tom Secunda, who is also chair of the Empire AI board, plus an initial $25 million research investment from SUNY. New York State provides operating capital, and computing support comes from Empire AI, Hochul’s $500 million public-private initiative. The release does not say how many tests the center will run, which models it will examine, or whether developers must give it access to their systems. Those details are not established by the sources available.

“The stakes are too high to simply take AI companies at their word.”

That was Hochul’s framing, alongside her statement that New York is acting “while the federal government fails to act.”

The testing arm joins a law that is about to bite

The center sits next to the RAISE Act, New York’s frontier AI law. DIGIT has opened an online RAISE Portal, and large frontier developers must register by January 1, 2027. From January they must publish safety frameworks on their websites. All frontier developers operating in New York must report critical safety incidents, with a 72-hour reporting deadline, and publish model transparency reports. The Department of Financial Services leads implementation through DIGIT, and Hochul says she will look at further ways to build on the law in the coming months.

The center is also the follow-through on an earlier promise. Hochul first announced an independent AI research center at Binghamton on January 21, 2026, at the time describing it as a public academic center for research on the safety, security and transparency of AI models. What is new now is the testing role, the named co-directors, the 80-researcher network and the formal link to RAISE oversight.

Why independent testing matters to developers and readers

So far, public information about frontier model safety has largely come from the developers themselves. A state-backed group that can check those claims changes who holds the evidence. For developers, the practical consequence is that the frameworks they must publish under RAISE will be reviewed by people with their own technical expertise.

Washington’s equivalent is under scrutiny. The Associated Press reported, per Techmeme’s summary, that officials at the Commerce Department’s AI safety center CAISSI were asked to report informally to a White House office, raising independence concerns. That is a report based on unnamed sources, not a confirmed finding, but it shows why a state-level body with its own funding and university home presents itself as independent.

Whether NYCAIRR can deliver depends on access. Its findings can only be as strong as the models and data it is allowed to examine, and the announcement does not describe any such arrangement. For now, those building or deploying AI in New York should expect the center to be advisory to the State, with the RAISE Act’s reporting and publication duties as the binding part.