# New York opens a state-run AI testing center at Binghamton University > Source: > Published: 2026-11-11 07:34:28+00:00 *New York's governor says the state won't take AI companies at their word anymore, and has built a lab to check. If you build or sell models, a regulator with its own test results is a different kind of counterpart from one holding your PDF.* Governor Kathy Hochul launched what her office calls the nation's first state-run AI testing program on Tuesday, November 10. According to Engadget, the work will happen at a new New York Center for AI Responsibility and Research at Binghamton University, with about 80 researchers involved and computing power supplied by Empire AI, the state's $500 million public-private AI partnership. Engadget quoted Hochul saying: "The stakes are too high to simply take AI companies at their word." The center's job is to test advanced AI systems, study their risks and hand public officials evidence they can act on. It will also look at responsible uses of AI in medicine and scientific discovery. Bloomberg Government carried the same announcement, describing it as the first public AI safety lab in the country. The limit is written into the design. As reported by Engadget, the program can only evaluate systems that companies have made public, so a model sitting unreleased inside OpenAI, Anthropic or Google DeepMind is out of reach. That's a real gap. The riskiest capabilities tend to show up in the weeks before a launch, not after it. Engadget's report does not say how results will be published, or which models go first. Hochul's office hasn't named companies. If you're waiting for a league table of labs, don't expect one on day one, and until the state says otherwise the safe assumption is that findings go to officials, not to the public. [New York becomes the first US state to ban new hyperscale data centers and other states are watching](https://startupfortune.com/new-york-becomes-the-first-us-state-to-ban-new-hyperscale-data-centers-and-other-states-are-watching/) New York Governor Kathy Hochul signed Executive Order 62 on July 14, making the state the first in the US to impose a moratorium on new hyperscale data centers drawing 50 megawatts or more. With over 300 data center bills filed across 30 states in 2026, the one-year pause gives other states a legal template to follow , and forces AI infrastructure... - [new york data center ban states](https://startupfortune.com/new-york-becomes-the-first-us-state-to-ban-new-hyperscale-data-centers-and-other-states-are-watching/) - [hyperscale data centers power grid impact](https://startupfortune.com/new-york-becomes-the-first-us-state-to-ban-new-hyperscale-data-centers-and-other-states-are-watching/) ## The law behind it The testing center sits alongside New York's RAISE Act, the frontier AI safety law Hochul signed in December 2025 and finalized with a chapter amendment on March 27, 2026, according to law-firm write-ups from Wiley and Cooley. The statute takes effect on January 1, 2027. Large developers have to register with the state by then, publish safety frameworks and transparency reports, and report critical safety incidents within 72 hours. The final version also creates an oversight office inside the Department of Financial Services. Read together, the two pieces fit. The law makes companies disclose; the lab gives the state a way to see whether what they disclose holds up when someone else runs the model. Frankly, that second half is what most state AI laws lack. A disclosure regime without anyone qualified to read the disclosures is paperwork. ## What it means if you ship models For the big labs, the practical change is small this year and larger from January. They already publish system cards and run their own red-team exercises. What's new is a government body with real compute that can rerun claims against the public product, and it can do that on its own schedule. Startups are the harder question. The RAISE Act is aimed at large developers, so a seed-stage team fine-tuning an open-weights model isn't the target. But a testing center that studies publicly available systems has no obvious reason to stop at the biggest names, and state lawmakers elsewhere are watching the model. If you sell into New York banks, hospitals or agencies, expect procurement teams to start citing the center's findings, whether or not you were ever formally covered. That is the compliance cost to plan for, and it comes less from a fine than from a buyer asking why your model scored the way it did. It also keeps the federal preemption fight alive. Washington has floated overriding state AI rules, and a state that runs its own lab makes a stronger case that it's doing something Congress hasn't. Nothing in the announcement says when the first results land. The next fixed date is January 1, 2027, when registration under the RAISE Act opens the books on the companies the lab will be looking at. **Also read:** [Cohere CEO Aidan Gomez says the Nvidia-backed lab still fights daily for AI chips](https://startupfortune.com/cohere-ceo-aidan-gomez-says-the-nvidia-backed-lab-still-fights-daily-for-ai-chips/) • [How to Calculate Token Cost Per Customer for AI Startup Pricing](https://startupfortune.com/how-to-calculate-token-cost-per-customer-for-ai-startup-pricing/) • [Big Walk's ten lines of AI code cost it 19 IGF jurors' votes](https://startupfortune.com/big-walks-ten-lines-of-ai-code-cost-it-19-igf-jurors-votes/) [US AI safety center staff now report informally to the White House, AP says](https://startupfortune.com/us-ai-safety-center-staff-now-report-informally-to-the-white-house-ap-says/) Commerce's CAISSI runs on a $10 million budget, about one-eighth of its UK counterpart's, has had two directors leave since April and has stopped publishing test results. Staff now answer informally to a White House office, the AP reports. - [AI safety center staff report to White House](https://startupfortune.com/us-ai-safety-center-staff-now-report-informally-to-the-white-house-ap-says/) - [did CAISSI stop publishing AI model test results](https://startupfortune.com/us-ai-safety-center-staff-now-report-informally-to-the-white-house-ap-says/) *This article is posted in [AI News](https://startupfortune.com/category/ai/), check it out for more related stories.* ## Join the discussion [Open in the community →](https://startupfortune.com/community/) Almost there. Sign in and your reply posts straight away.