Microsoft launches its first cybersecurity model and a new agentic system
Microsoft has announced the launch of its first AI model specialized in cybersecurity, MAI-Cyber-1-Flash, and a new agentic security platform called Perception, aiming to compete with rivals like Anthropic, Google, and OpenAI.
On Monday, at a small event in San Francisco, Microsoft unveiled its first cybersecurity-specialized AI model alongside a new AI-powered security platform, taking a direct aim at major competitors such as Anthropic, Google, and OpenAI.
The model, named MAI-Cyber-1-Flash, is designed to find challenging vulnerabilities in complex codebases. It is built to work with MDASH, Microsoft’s harness for software vulnerability identification and remediation.
The new security platform, called Perception, is designed to deploy teams of agents to assist with and automate various security workflows, including identifying and fixing bugs. It can also integrate with MDASH.
Microsoft claims that MAI-Cyber-1-Flash is significantly more powerful and cost-effective than competitor models, based on its performance on the Cyber Gym benchmark, a recognized AI cybersecurity standard.
“We’re very, very excited to announce our results,” said Mustafa Suleyman, co-founder of DeepMind and current CEO of Microsoft AI. “We have MAI-1 Cyber Flash binded with GPT 5.4 inside the MDASH harness — which beats out Gemini, GPT 5.5 Cyber, GPT 5.6 Sol, and Mythos 5 on Cyber Gym, the golden benchmark.”
“We’re shipping this into production immediately,” he added.
Noting that hackers are increasingly using AI in their cyberattacks, Hayete Gallot, Microsoft’s vice president for security, described Perception as a way for enterprise defenders to “defend against AI with AI at the scale and speed that the attackers have.”
Perception uses agentic red teams, blue teams, and green teams. Red teams can provide detailed simulations of potential attacks, offering context about threat actors and vulnerabilities. Blue teams are dedicated to detecting and triaging existing bugs, while green teams take corrective actions.
Dave Weston, lead engineer for Perception, described the platform as a massive efficiency upgrade. “We’ve gone from this taking hours and hours of manual work from multiple specialized folks across the security organization — and in minutes, we have a fix for all of this. Not only do we discover the issues and prioritize them, but we have detection, posture fixing, and even a code fix.”
While AI has provided new defensive capabilities, its availability to cybercriminals has created a wide array of potential threats. Microsoft’s new security tools, available in preview from November 3, will enter an increasingly crowded field of AI cybersecurity solutions. Earlier this year, Anthropic launched Mythos, and OpenAI released its own security solution in May.
