Mass General Brigham Won't Let AI Agents Act Alone Yet

Mass General Brigham is drawing a hard line on how far it will let artificial intelligence go on its own. The Somerville, Massachusetts-based system has spent three years building AI governance, and its chief technology officer told Becker's Hospital Review that the arrival of agentic AI, tools that can take action rather than just generate text, is pushing her toward a more restrictive posture sooner than she anticipated.

The distinction matters. Generative tools draft notes or summarize records, and a clinician reviews the output. Agentic tools can execute tasks, such as scheduling, ordering, or moving data, without a human checking each step. That autonomy raises the stakes on errors, accountability, and patient safety, so Mass General Brigham wants humans firmly in the loop before agents run independently.

The caution reflects a broader tension across health systems racing to adopt AI while their oversight frameworks lag behind the technology. As vendors push agentic features, hospital leaders are being forced to decide how much control to cede, and one of the country's largest academic systems is signaling restraint.