Microsoft is reportedly preparing to launch a new AI cybersecurity product designed to identify software vulnerabilities and automatically fix them, as the company expands its push into AI-driven enterprise security.
According to The Information, the product, internally known as Project Perception, could debut as early as this month. It combines AI models from Microsoft, OpenAI, and Anthropic via a model-routing system that selects the most suitable model for each task, allowing Microsoft to reduce costs while offering capabilities similar to Anthropic's cybersecurity-focused Mythos model. The report said the product is intended to help enterprises detect software bugs and remediate vulnerabilities automatically while providing a more cost-effective alternative to Anthropic's offering.
The launch comes as enterprises increase spending on AI-assisted cybersecurity tools amid growing concerns that attackers are also adopting generative AI to accelerate the discovery of software flaws. Project Perception is reportedly also part of a broader overhaul led by Microsoft's new security chief, Hayete Gallot, who has been reshaping the company's security business around AI-powered products.
Microsoft has already laid the groundwork for AI-based security through its Copilot for Security platform. According to Microsoft's official announcements, the service uses generative AI to help security analysts investigate incidents, summarize threats and accelerate response times, while integrating with Microsoft Defender, Sentinel, and other security products.
Article edited by Jack Wu
