Binghamton gathers AI safety experts to shape New York research center’s agenda
Organizers will turn the discussions into initial research plans, with agent containment and academia’s role as a check on industry among the issues raised.
Loading page…
Organizers will turn the discussions into initial research plans, with agent containment and academia’s role as a check on industry among the issues raised.
Listen to this story
The New York Center for AI Responsibility and Research is moving from launch toward defining its first research program: organizers will synthesize ideas from Binghamton’s safety conference and plan the center’s initial rollout with its interim co-directors. The agenda is not yet settled, but the center is expected to draw on more than 80 researchers statewide. Its work will have access to Empire AI computing support, distinct from the center’s own funding.
Governor Kathy Hochul appointed Greg Morrisett and Jeannette Wing interim co-directors on November 10, two days before the conference.
The center was announced with a $30 million philanthropic gift; SUNY added an initial $25 million research investment, and New York state provided operating capital.
Morrisett highlighted research on keeping AI agents contained, including securing sandboxes and the software harnesses around them.
New York’s new AI research center has a statewide safety mission, but its initial research agenda is still taking shape. Binghamton University’s November 12–13, 2026, conference brought together academic, industry and government leaders to help define that work. Organizers now plan to turn the discussions into the center’s first efforts, the university said November 18.
The Research Frontiers for AI Safety and Responsibility conference covered safety, verification, transparency and hardware verifiability. Participants also had opportunities to connect with potential research partners and industry leaders. Organizers hope those exchanges will inform a research roadmap for the New York Center for AI Responsibility and Research, hosted at Binghamton.
Kuang-Ching “KC” Wang, a forum organizer and director of Binghamton’s School of Computing, described the next step as synthesizing the discussions. Organizers will then work with the center’s interim co-directors to plan its initial rollout. That places the conference’s output at the planning stage, rather than presenting it as a settled research program.
The leadership was appointed just before the gathering. Governor Kathy Hochul named Cornell Tech’s Greg Morrisett and Columbia University’s Jeannette Wing interim co-directors on November 10. They are expected to bring together more than 80 scientists from New York colleges and research institutions, drawing on expertise in AI safety, security and computer science.
The center itself was announced in January with a $30 million philanthropic gift from donors led by Bloomberg LP co-founder Tom Secunda. An initial $25 million research investment from SUNY and operating capital from New York state accompanied that gift. Its computing support comes from Empire AI, the state’s $500 million public-private AI research initiative—not a separate $500 million budget for the center.
Morrisett illustrated the breadth of the assignment with concerns he has heard from two very different people. A mother wanted to know what AI would mean for her children’s education and whether interacting with it was safe. A bank official deploying an AI agent wanted assurance that accounts would remain secure. Morrisett argued that addressing such questions requires a large research network.
He and Wing see many AI problems as security problems. One example is an agent breaking out of a sandbox: a contained environment intended to keep it away from the wider internet. Morrisett called for securing both those environments and the surrounding software controls, or harnesses, so agents cannot take harmful actions.
His proposed approach draws on existing computer-security methods. Researchers can use many of the same systems techniques that contain malicious software to contain agents, he said. He presented that containment work as an important step for the center’s field.
It's not good just to wait and see what industry does and hope it turns out well for all of us. I don't think that's a good way to go. It's not a responsible way to go.
Renzi Arpaci-Dusseau, founding dean of the University of Wisconsin–Madison’s College of Computing and Artificial Intelligence
Keynote speaker Renzi Arpaci-Dusseau made a broader argument about who should shape the technology. The University of Wisconsin–Madison computing dean described academia as a counterweight to technology companies and their profit motives. His call was for universities to participate in the changes AI brings, rather than leave their direction to industry.
That argument also acknowledged the appeal of AI-assisted software creation. Arpaci-Dusseau described a shift from relatively static code, written by professionals, toward software people build for themselves. He called that change “liquid infrastructure” and said the ability to create personal software was exciting. His position paired enthusiasm for that new freedom with a responsibility for universities to help shape its development.
Loading discussion...
Join the conversation
Explain where academic responsibility should begin and end.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.