New York will have its own scientists vet AI models instead of taking the companies' word for it
A new centre at Binghamton University with more than 80 researchers is to help the state oversee large AI developers, although it gets no access to unreleased models.
Governor Kathy Hochul on Tuesday presented the New York Center for AI Responsibility and Research in Albany, the State of New York itself announced. The centre is based at Binghamton University and brings together more than 80 scientists from colleges and universities across the state, who according to Finger Lakes 1 will keep working at their own institutions. The interim directors are Greg Morrisett of Cornell Tech and Jeannette Wing of Columbia University.
The centre is to test advanced AI systems independently for safety, develop measurement methods and standards and give policymakers evidence-based advice. To do so it will use computing power from Empire AI, a 500 million dollar public-private initiative. There are 30 million dollars in donations, led by Bloomberg co-founder Tom Secunda, and 25 million dollars from the state university system SUNY. Hochul justified the step by arguing that too much is at stake to trust AI companies blindly.
The centre will also assess the safety frameworks that large AI developers must publish under the RAISE Act. The law requires them to register by 1 January 2027, publish transparency reports and report serious incidents within 72 hours. Engadget does point to a limitation: the centre can only work with what companies make public and cannot examine models that have not yet been released. According to Engadget, New York is also positioning itself against Washington, where the White House proposed a framework that would override state laws.