Microsoft Unveils Its Pioneering Self-Developed Secure Large Model, MAI-Cyber-1-Flash, Outshining Rivals in System Benchmark Tests MDASH
2 day ago / Read about 0 minute
Author:小编   

On July 28, 2026 (local time), Microsoft proudly announced the launch of its inaugural self-developed cybersecurity AI model, MAI-Cyber-1-Flash, accompanied by an agent-based security system named Project Perception. MAI-Cyber-1-Flash is specifically engineered to pinpoint security loopholes within intricate codebases. When seamlessly integrated with Microsoft's Secure Multi-Model Agent Scanning Platform, it delivers comparable vulnerability detection prowess at a mere fraction—half—of the cost incurred by leading models. Project Perception elevates AI-assisted cybersecurity defense by perpetually monitoring, rectifying, and sealing off emerging threat vectors through a coordinated effort of multiple agent groups.