Follow the latest coverage, related explainers and connected technology stories.
Microsoft reviews the results of the FORGE Lab in using multi-model agentic systems to discover and validate vulnerabilities across Windows, the Linux kernel, and open-source projects. It concludes that the challenge is no longer merely finding flaws, but building a pipeline capable of validating and fixing them and incorporating them into the release cycle.
Microsoft warns that advanced artificial intelligence models will increase the volume of discovered vulnerabilities and the speed at which they are exploited, requiring chief information security officers to improve triage, remediation, and defense in depth. The company recommends shortening remediation windows for critical systems, using model control layers, and strengthening secure-by-default settings.
Microsoft has unveiled Project Perception, an agentic security system designed to monitor and analyze risks and take defensive action at machine speed while keeping humans in control. The system enters public preview on August 3 and uses a multi-model architecture that includes MAI-Cyber-1-Flash for software vulnerability management.