Microsoft is preparing to launch a new product using artificial intelligence to automatically detect and fix software vulnerabilities, directly challenging Anthropic's cybersecurity model "Claude Mitos." According to a report published on the 16th (local time) by U.S. tech publication The Information, the forthcoming product will simultaneously leverage multiple AI models to analyze software code and identify bugs, then automatically generate remediation code. The core technology is believed to be "MDASH," Microsoft's internally developed AI-based vulnerability detection system, which combines high-performance reasoning models with cost-efficient lightweight models. Microsoft previously disclosed that it used MDASH to discover 16 new vulnerabilities in Windows network and authentication systems. Mitos is currently available only on a limited basis to select vetted enterprises and institutions. The product launch coincides with a restructuring of Microsoft's security business unit. Hayete Gallot, Corporate Vice President overseeing Microsoft's security division, has replaced at least nine Corporate Vice President-level executives in roughly five months since assuming her role in February, and laid off several hundred employees from the division's total workforce of approximately 10,000. Teams for traditional security products were scaled back, while AI security product teams were expanded.