I recently wrote about why we need to treat models as insider risks, with security controls sitting outside the model itself. [Read more]

Microsoft expanded Project Perception, its agentic security system announced in July, with new capabilities to continuously assess agent risk, adapt permissions and contain threats from AI agents deployed inside enterprises. Microsoft CEO Satya Nadella framed the effort as applying "deterministic controls on non-deterministic intelligence" to what he called the "AI Insider" — agents that often carry more privileges than they need and are trained to get results rather than follow organizational norms. The expansion extends the cyber stack Microsoft introduced in July, which uses red, blue and green agent teams to defend against AI-driven attacks from outside the enterprise.

I recently wrote about why we need to treat models as insider risks, with security controls sitting outside the model itself. Today, were putting that principle into practice by expanding Project Perception with new capabilities to help continuously assess agent risk, adapt permissions, and contain threats. The emergence of powerful cyber models has driven organizations to focus on attackers leveraging AI to run their campaigns. We must also address the agents already deployed inside our walls: the AI Insider. Agents are now running across the enterprise at scale. They often have more privileges than they need. They aren't malicious. But they're trained to get results, not to follow an organizations norms or judgment. Securing the AI insider requires protection across multiple layers: identity, permissions, data, endpoints, network, cloud, and runtime. In July, we announced Perception, an agentic security system that uses our new cyber stack and teams of red, blue, and green agents to help defenders act with speed, while defending against AI-driven attacks from the outside of the enterprise. Today, we are extending that cyber stack to help defenders secure AI Insiders by implementing deterministic controls on non-deterministic intelligence. Agentic security provides speed. Practitioners provide judgment. Learn how Perception can help defenders operate at the same speed as the systems they are protecting against. There shouldn't be a choice between innovation and safety. https://lnkd.in/geZZSXhd https://lnkd.in/geZZSXhd?trk=public post reshare-text See more comments https://www.linkedin.com/signup/cold-join?session redirect=https%3A%2F%2Fwww%2Elinkedin%2Ecom%2Fposts%2Fsatyanadella securing-the-agentic-insider-the-official-activity-7517971322266640384-ahKI&trk=public post see-more-comments