Microsoft launches MAI-Cyber-1-Flash cybersecurity model
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Microsoft launched MAI-Cyber-1-Flash, its first cybersecurity-specialized model, at a San Francisco event, claiming it beats Gemini, GPT 5.5 Cyber, GPT 5.6 Sol, and Mythos 5 on the Cyber Gym benchmark — which Suleyman called "the golden benchmark."
- Mustafa Suleyman, CEO of Microsoft AI and DeepMind co-founder, said the model ships into production immediately and is integrated with MDASH, Microsoft's harness for software vulnerability identification and remediation.
- Microsoft also unveiled Perception, an agentic cybersecurity platform deploying red, blue, and green teams for attack simulation, bug triage, and remediation — with lead engineer Dave Weston saying it collapses hours of manual security work into minutes.
- Perception integrates with MDASH and will be available in preview on November 3, entering a market where Anthropic launched Mythos (via its Glasswing partner program) earlier this year and OpenAI launched its Daybreak security solution in May.
- Hayete Gallot, Microsoft's VP of security, framed Perception as a way for enterprise defenders to "defend against AI with AI at the scale and speed that the attackers have," citing the rise of AI-powered cyberattacks.
Why it matters: Microsoft is making a direct competitive play in AI cybersecurity, publicly claiming benchmark superiority over Anthropic's Mythos and OpenAI's Daybreak — the two named rivals — with a November 3 preview that puts pressure on competitors still gating access behind partner programs like Glasswing.
