New York is launching a state-backed AI safety research center at Binghamton University to independently evaluate AI systems, develop safety and security techniques, and train researchers in responsible AI.

The center has $55 million in announced funding: a $30 million philanthropic gift led by Tom Secunda and a $25 million research investment from SUNY and New York State; the state will also cover operating costs.

Greg Morrisett and Jeannette Wing will serve as interim co-directors, setting the scientific agenda and overseeing independent evaluations while researchers remain at their home institutions.

The research will focus on preventing harm, anticipating failures and vulnerabilities, and improving defenses against attacks, including risks posed by autonomous agents.

Initial research and recruiting priorities include formal methods and verification, agentic and complex systems, and interpretability and alignment.

At the state’s request, the center can provide technical evidence for policymaking and work with the Office of Digital Innovation, Governance, Integrity, and Trust (DIGIT) to review developers’ published safety frameworks and recommend stronger requirements; DIGIT will administer registration and reporting.

A major limitation is that researchers will generally be able to assess only information companies make public, leaving unreleased and privately developed models out of view unless access is voluntarily provided or regulators compel disclosure.

The center will conduct research and provide evidence, but it has no automatic enforcement authority, and New York has not specified whether its findings will affect licensing, procurement, enforcement, or disclosure requirements.

Under amended RAISE Act provisions scheduled to take effect on Jan. 1, 2027, large frontier AI developers must register through the RAISE Portal; the Act defines covered developers as those with more than $500 million in annual gross revenue, including affiliate revenue.

The Act allows civil penalties of up to $1 million for a first specified violation and $3 million for subsequent violations.

Hochul said independent testing is essential because the stakes are too high to rely only on companies’ claims, and warned that competitive pressure could encourage developers to move too quickly.

The initiative is intended to build New York’s technical capacity for AI regulation as federal policy remains unsettled and states including New York and California pursue binding requirements.