IT Home July 17 news, citing a report by The Information, that Microsoft is developing an AI-based vulnerability detection tool, similar to Anthropic’s Mythos model.

Insiders revealed that the product is called “Project Perception,” a brand-new security offering designed to use AI to help enterprises identify and fix software vulnerabilities.

The product could be released as early as this month. It is expected to integrate AI models from Microsoft, OpenAI, and Anthropic, enabling it to handle a wide range of security tasks while effectively controlling operating costs.

Unlike security tools that rely on a single AI model, reports say that the “Perception Project” will use a technology known as “model routing” (model router). Before assigning workloads, the system will first assess and decide which AI model is best suited to handle a specific task. The platform can flexibly switch between different models, enabling tasks such as code analysis, identification of security weaknesses, and automatic generation of remediation plans.

Another goal of this strategy is to reduce costs. Anthropic’s AI model for cybersecurity, Mythos, has drawn widespread attention in the industry thanks to its advanced vulnerability discovery capabilities, but it is generally believed that the cost of deploying it at large scale is too high. IT Home learned that Microsoft hopes to distribute requests across multiple models rather than relying solely on a single model, so as to offer customers the same level of security capability at a lower price.

It is reported that the “Perception Project” is one of Microsoft’s first major initiatives led by its newly appointed security chief, Hayete Gallot, who officially took charge of the department this February. Since assuming the role, Gallot has reorganized Microsoft’s security business architecture, shifting the focus further toward AI-driven products while cutting investment in traditional ones.