New York recruits 80-plus scientists to test AI models itself

Governor Kathy Hochul rolled out the nation's first state-run AI testing program on Tuesday, putting independent researchers — not the AI companies themselves — in charge of checking what the technology actually does. "The stakes are too high to simply take AI companies at their word," Hochul said in a press release.

The New York Center for AI Responsibility and Research at Binghamton University will run the effort, with more than 80 New York scientists contributing. Computing power comes from Empire AI, the state's $500 million public-private AI partnership. Hochul's office said the center will test advanced AI systems, study their risks and supply evidence to inform decisions by public officials, plus look at responsible uses in medicine and scientific discovery.

The program has one obvious blind spot: it can only study what AI companies choose to disclose publicly. New York's RAISE Act, which Hochul signed last year, requires large AI developers to register with the state by January 1, 2027. From then on, those companies must publish safety frameworks and transparency reports, and flag critical safety incidents within 72 hours.

The law won't crack open unreleased models, but it does force developers to hand over more than they currently do — giving the Binghamton researchers a bigger surface to test against.