Key Takeaways

• Microsoft Research and Carnegie Mellon University’s AI Measurement Science & Engineering Center (AIMSEC) have announced an in-person AI agent evaluation and governance workshop for October 22–23, 2026.
• The event is planned for Microsoft Research New York City Lab at 300 Lafayette St, New York, NY 10012, with sessions scheduled from 9:00 AM to 5:00 PM.
• The organizers expect 120 participants drawn from academia, industry, civil society, and government.
• The program is set to include keynotes, lightning talks, panels, hands-on working groups, and breakout discussions.
• Discussions are intended to inform open public synthesis reports; the event has not yet announced finished standards or workshop findings.

What Microsoft Research and CMU Announced

Microsoft Research and Carnegie Mellon University’s AI Measurement Science & Engineering Center, known as AIMSEC, plan to convene 120 leaders for a two-day workshop on evaluating and governing AI agents. The official event listing dates the gathering for October 22–23, 2026, in New York City. The stated focus is developing shared standards and measurement science for AI agent evaluation and governance.

The announcement comes as AI agents move from research prototypes toward systems that act on people’s behalf. Microsoft Research identifies a gap between the pace of agent development and the ability to evaluate these systems. Its event description presents rigorous, shared measurement science as a foundation for accountability and says that evaluation remains a significant challenge for the field.

The workshop is an effort to bring relevant communities into a shared discussion. The announced participants are expected to include people from academia, industry, civil society, and government, alongside team members from Microsoft and Carnegie Mellon. The source does not provide a complete attendee list, name individual participants, or specify how places will be allocated.

This is a convening, not a product announcement. The official information does not announce a new AI model, commercial service, benchmark score, or completed evaluation standard. Its purpose is to bring participants together to address the science and practice of evaluating and governing AI agents.

Event Details at a Glance

These details are drawn from the Microsoft Research event listing. The announcement identifies the event as in person in New York; it does not state that a remote attendance option is available.

What the Workshop Will Cover

The event description groups its aims around convening people who can contribute to evaluation standards and surfacing the state of the art. The planned program combines keynote presentations, lightning...