Microsoft Research Blog
Red-teaming a network of agents: Understanding what breaks when AI agents interact at scale
Safe agents don’t guarantee a safe ecosystem of interconnected agents. Microsoft Research examines what breaks when AI agents interact and why network-level risks require new approaches.
Publication
Diagnosing Capability Gaps in Fine-Tuning Data
Career Opportunity
Principal Applied Scientist – Microsoft Copilot
Conversational commerce introduces challenges that differ from traditional web shopping. Preferences emerge through dialogue, expectations for accuracy and trust are high, and systems must reason over context and frequently changing commerce data. Microsoft Copilot is…
Career Opportunity
Senior Applied Scientist – Crowd Intelligence Platform (CIP) team
The Crowd Intelligence Platform (CIP) team is looking for a Senior Applied Scientist to help design, build, and scale AI‑powered testing systems that can be applied generically across a wide range of cutting‑edge MAI products.…