Microsoft recently officially launched its first specialized cybersecurity model, MAI-Cyber-1-Flash, and a new AI cybersecurity platform called Perception, during an event in San Francisco, further strengthening its position in the field of artificial intelligence security and competing with companies such as Anthropic, Google, and OpenAI.
Microsoft stated that MAI-Cyber-1-Flash is a model specifically designed for cybersecurity scenarios, aiming to identify complex vulnerabilities in codebases and provide core capabilities to Microsoft's vulnerability identification and remediation tool MDASH. Combined with the Perception platform, this model can help enterprises automate security processes such as vulnerability discovery, analysis, and remediation.
Mustafa Suleyman, Microsoft's AI CEO, said that the company has integrated MAI-1Cyber Flash with GPT5.4 within the MDASH framework and outperformed competing models such as Gemini, GPT5.5Cyber, GPT5.6Sol, and Mythos5 in the Cyber Gym cybersecurity benchmark test. Microsoft claims that the model offers advantages in both performance and cost efficiency and plans to immediately deploy it in production environments.
At the same time, the Perception platform introduced by Microsoft uses a multi-agent collaboration architecture, consisting of three types of AI agents: red team, blue team, and green team. The red team agent simulates potential attack behaviors and analyzes threat paths; the blue team agent is used to detect and classify vulnerabilities; and the green team agent performs vulnerability remediation and security optimization tasks.
Hayete Gallot, Microsoft's Vice President of Security, stated that as hackers increasingly use AI to launch attacks, companies need to leverage AI to improve defense efficiency. Perception aims to help security teams use AI for protection at speeds and scales comparable to attackers.
David Weston, Chief Engineer of Perception, said that previously, security teams needed multiple professionals to spend hours completing vulnerability detection and remediation, but with AI agent collaboration, enterprises can now complete vulnerability discovery, prioritization, security testing, and code fixes within minutes.
