Shenoy then offered a perspective grounded in real-world deployment. He described his institution—a research medical school in the Bronx with more than $250 million in research funding—as “right on the cusp of danger.”
Human identity is well-managed. NHI is the new frontier, and the stakes are high. An AI agent that goes unchecked in a research lab can quietly become an AI agent making decisions about patient data—without anyone able to trace how it got there.
“AI agents are like employees, like research assistants,” he said. “There should be accountability that goes up to the lab head, the head of the business. We don’t have that today.”
That’s why he began working with IBM. Today, Albert Einstein uses IBM Verify and HashiCorp Vault together—Verify for the human side, Vault for the application side—to bring NHIs under coordinated control before agentic AI scales further across the institution.
His advice to other leaders was direct: “This is something we have to address—no head in the sand. I’m not going to wait for someone to ask me. I’m not going to wait for a regulator. This is the proper hygiene, and it’s what we need to do.”
To dive deeper, watch Bob Kalka and Tyler Lynch’s on-demand webinar, “Agentic IAM in Practice: Enforcing Least Privilege and Auditability for AI Agents.”