Microsoft bolsters security to better control AI insider risks

[French headquarters of Microsoft, Issy-les-Moulineaux, France] HJBC/iStock Editorial via Getty Images Microsoft (MSFT [https://seekingalpha.com/symbol/MSFT]) has expanded on the capabilities of Project Perception, an agentic security system that uses its new cyber stack, to protect against insider attacks as well. While originally released in July to defend against outsider AI-powered attacks, the risks associated with enterprises' internally released agents have prompted the need for security against what Microsoft calls "agentic insiders." "This is not because models are inherently malicious," said Microsoft's executive vice president of security Hayete Gallot in a Monday blog post.