Microsoft has unveiled its first cybersecurity model and an agentic platform.

On Monday, at an event in San Francisco, Microsoft unveiled its first specialized model for cybersecurity, along with a new AI-powered protection platform, thereby challenging Anthropic, Google, and OpenAI.

MAI-Cyber-1-Flash is a model, according to the company's description, designed to identify complex vulnerabilities in large codebases. It operates within MDASH, Microsoft's framework for detecting and remediating software vulnerabilities.

Perception is a new security platform that deploys agent teams to automate cybersecurity tasks—such as detecting and fixing bugs—and integrates with MDASH. It utilizes three types of agent teams:

— red team: simulates potential attacks by assessing likely threat actors and the vulnerabilities they might exploit

— blue team: focuses on detecting and prioritizing existing bugs

— green team: takes action to remediate them.

Hayet Gallou, Microsoft's Vice President of Security, highlighted the growing use of AI by hackers and described Perception as a way for corporate defenders to counter AI-driven attacks using AI—at the same speed and scale as the attackers.

Microsoft's new tools will enter preview access on November 3, entering an increasingly competitive market for AI-powered cybersecurity solutions. Earlier this year, Anthropic introduced the Mythos platform, available to a select group of partner organizations through its Glasswing program, while in May, OpenAI launched a similar solution via its Daybreak program.

Source: microsoft.com