Microsoft is building 'Project Perception', a new bug-finding AI tool

Microsoft is working on a new AI security tool called Project Perception. It helps companies find and fix software bugs. The product could launch as soon as this month, and it will use AI models from Microsoft, OpenAI, and Anthropic together.

Unlike tools that rely on just one AI model, Project Perception uses a 'model router' system. This system picks the best AI for each job—like code analysis, spotting security flaws, or creating fixes. By spreading tasks across multiple models, Microsoft hopes to keep costs down instead of depending only on expensive models like Anthropic's Mythos.

The project is led by Microsoft's new security chief, Hayete Gallot. She took over in February and has been shifting the security division's focus toward AI-powered products while cutting back on traditional ones. This shows Microsoft is pushing AI into cybersecurity to be more efficient and competitive.