Microsoft launches its first cybersecurity model, as well as a new cybersecurity system

Microsoft on Monday launched its first cybersecurity prototype alongside a new AI-powered cybersecurity platform at a small event in San Francisco, drawing significant criticism from the major players in the space — Anthropic, Google and OpenAI.

The company describes MAI-Cyber-1-Flash as a model designed to “find difficult vulnerabilities in complex code bases.” The model is designed to drive MDASH, Microsoft’s tool for identifying and remediating software vulnerabilities.

The new security platform is called Perception, and is designed to deploy teams of agents to help and automate various security workflows, including identifying and remediating errors. The platform can also integrate with MDASH.

The company claims that the MAI-Cyber-1-Flash is significantly more powerful (and more cost-effective) than competing models, based on its performance under an AI-based cybersecurity standard.

“We are very excited to announce our results,” said Mustafa Soliman, co-founder of DeepMind and current CEO of Microsoft AI. “We have an MAI-1 Cyber ​​Flash linked [sic] With GPT 5.4 inside the MDASH belt – which beats Gemini, GPT 5.5 Cyber, GPT 5.6 Sol and Mythos 5 on the Cyber ​​Gym, which is the basic benchmark we all use. The gold standard.”

“We are shipping this product into production immediately,” he added.

Noting that hackers are increasingly using AI in their cyberattacks, Haiti Goliath, Microsoft’s vice president of security, described Cognition as a way for enterprise defenders to “defend against AI using AI at the scale and speed that attackers have.”

Perception uses proxy red teams, blue teams, and green teams. Red teams can provide detailed simulations of potential attacks, providing context about potential threat actors and potential vulnerabilities they may exploit. Blue teams are dedicated to detecting and triaging existing errors, while green teams take “corrective actions” against those errors.

Dave Weston, lead architect for Perception, described the platform as a huge efficiency upgrade for corporate advocates. “It took hours and hours of manual work from many specialized people across the security organization — application hunters, remediation engineers, you name it — and in minutes, we have a fix for all of this. Not only do we detect and prioritize issues, we have detection, fix the situation, and even fix the code.”

Although AI has provided new defensive capabilities for businesses, its availability to cybercriminals has created a dizzying array of potential threats.

Microsoft’s new security tools, which the company said will be available for preview on November 3, will enter an increasingly crowded field of AI-based cybersecurity solutions. Earlier this year, Anthropic launched Mythos, a security platform released to a small group of partner organizations through a program called Glasswing. OpenAI also launched its own security solution in May through a program called Day Break.