Microsoft has announced two major cybersecurity advancements: its first in-house security AI model, MAI-Cyber-1-Flash, and Project Perception, an agentic AI security system.
Microsoft integrated MAI-Cyber-1-Flash into MDASH, achieving 95.95% accuracy on CyberGym while cutting operational costs 50%. Specialized compact model handling 90% of workloads plus frontier GPT-5.4 for complex tasks outperforms general-purpose alternatives at half the cost.
When you purchase through links on our site, we may earn an affiliate commission. Here’s how it works.
Microsoft's new security setup achieves a 95.95% success rate on the CyberGym benchmark while cutting operational costs by nearly 50%.
Pradeep Viswanathan
Neowin
@pradeepviswav ·
Microsoft launches its own cybersecurity model MAI-Cyber-1-Flash but still depends on OpenAI for the toughest tasks
Microsoft unveils AI security tools it says outperform competing platforms
Microsoft launches its first cybersecurity model, plus a new agentic cybersecurity system | TechCrunch
Microsoft's MDASH outperforms Claude Mythos, GPT-5.6 in cybersecurity test
Microsoft touts cost-saving AI model for cybersecurity
Microsoft's first cybersecurity model powers new Project Perception agents - SiliconANGLE
Microsoft Says MDASH Beats Claude Mythos and GPT-5.6 Sol in Cybersecurity Test - Decrypt
Microsoft launches AI cybersecurity model, agentic defense platform to cut enterprise security costs
Microsoft's solution to AI security: more AI and more acronyms
Microsoft built an agentic security system with red, blue, and green team AI agents. It enters public preview August 3.
