agentic security systemand a specialized AI model designed to analyze software vulnerabilities at scale.
Microsoft Releases MAI-Cyber-1-Flash
On Monday, Microsoft introduced MAI-Cyber-1-Flash, its first security-focused AI model. Based on Microsoft AI’s internally developed MAI-Thinking-1 reasoning model, it integrates into the MDASH bug-hunting harness.
Microsoft executives claim the model, boosted by GPT-5.4, outperforms rival AI systems while cutting operational costs by approximately half.
Performance Benchmarks Show Clear Lead
CyberGym’s benchmarking revealed the performance gap. The MAI-Cyber-1-Flash and GPT-5.4 combination within the MDASH harness achieved a 95.95 percent success rate in vulnerability detection and patching.
| AI System | Success Rate |
|---|---|
| Microsoft MAI-Cyber-1-Flash + GPT-5.4 | 95.95% |
| OpenAI GPT-5.5 Cyber | 85.6% |
| OpenAI GPT-5.6 Sol | 83.6% |
| Anthropic Mythos 5 | 83.8% |
| Google Gemini 3.5 Flash Cyber | 83.2% |
Mustafa Suleyman, CEO of Microsoft AI, called this a remarkable result.
He explained that MAI-Cyber-1-Flash handles 90 percent of vulnerability queries, detecting and patching issues automatically. The larger GPT-5.4 model tackles the remaining 10 percent of complex cases.
Project Perception Orchestrates Coordinated Defense
Microsoft also announced Project Perception, an agentic security system that coordinates three types of AI agents working in concert.
- Red team agents find vulnerabilities and simulate attack paths
- Blue team agents investigate identified risks and their potential impact
- Green team agents remediate issues and patch systems
Hayete Gallot, EVP of Microsoft Security, emphasized the need for defenders to operate at the speed and scale of attackers. She stated: You need a new cyber stack. So we built it. This is what we call Perception.
Expanding AI Security Research Across Universities
Microsoft also introduced Security FORGE (Frontier Offensive Research and Generative Exploration) Labs, a new AI security research arm. Taesoo Kim, Microsoft VP of Security Research, leads the initiative.
Additionally, the External Red Team Alliance (EXTRA) aims to expand AI safety research globally. Microsoft has provided unrestricted funding to 18 university labs across six continents to accelerate this work.
The Coming Wave of Patches
AI is discovering countless previously hidden vulnerabilities at an unprecedented pace. This signals what security teams describe as a hot, messy summer
ahead.
Microsoft itself has already warned customers that AI-powered vulnerability discovery will mean significantly busier Patch Tuesdays. What once happened on a predictable schedule is becoming a constant stream of critical updates.
The implication is clear: the era of leisurely patch cycles is over. Security teams must be prepared to move faster than ever before.
Follow Hashlytics on Bluesky, LinkedIn, Telegram and X to Get Instant Updates
