Back Feeds.4Sysops

Microsoft pairs Project Perception with runtime defenses for AI agents

Feeds.4Sysops •IT News • July 28, 2026

Microsoft is pairing Project Perception’s autonomous cyberdefense with Microsoft Defender protections that can detect and block risky AI agent behavior at runtime. Project Perception enters public preview August 3, while local agent runtime protection on Windows is already available in public preview. Source

The full story

This article is one source in a clustered incident — the cluster page carries the summary, timeline and every other outlet covering it.

Microsoft Unveils Project Perception to Combat AI-Powered Cyber Threats On July 27, 2026, Microsoft launched Project Perception, an AI-driven security platform designed to counteract the growing threat of AI-based cyber attacks. This system utilizes three types of specialized agents—red, blue, and green—to identify, assess, and remediate vulnerabilities in real-time. The platform integrates the new MAI-Cyber-1-Flash model, which can handle 90% of tasks, while the remaining 10% are managed by the more powerful GPT-5.4 model. Microsoft claims that this combination achieved a 95.95% success rate in the CyberGym benchmark, outperforming competitors while reducing operational costs by nearly 50%. Project Perception will enter public preview on August 3, 2026, and is designed to operate in environments without internet access, ensuring data security and privacy. The initiative reflects a significant shift in cybersecurity, addressing the rapid evolution of autonomous AI threats. 7 sources 58

Extracted Entities

Platforms (2)

Microsoft Defender Windows