Nadella Expands Project Perception to Contain Agent Risk

Satya Nadella wants agent security to sit outside the model.

“I recently wrote about why we need to treat models as insider risks, with security controls sitting outside the model itself,” @satyanadella posted Monday morning.

“Today, we’re putting that principle into practice by expanding Project Perception with new capabilities to help continuously assess agent risk, adapt permissions, and contain threats.”

The same concern runs through this edition. OpenCode says AI models helping it fight an ongoing cyberattack began refusing mid-incident, and Ethan Mollick warns against giving today’s agents control of your bank account.