Microsoft Unveils MAI-Cyber-1-Flash Cybersecurity AI Model
Summary
Microsoft says MAI-Cyber-1-Flash, integrated with its MDASH vulnerability-finding system, performed strongly in CyberGym testing. The model will be offered through Project Perception, scheduled to enter public preview on August 3.
Key points
- Microsoft says MAI-Cyber-1-Flash is designed to identify difficult vulnerabilities in complex code and is integrated into its MDASH system.
- MDASH orchestrates more than 100 specialized AI agents across multiple models and has been used to find vulnerabilities in Microsoft codebases.
- Microsoft says MAI-Cyber-1-Flash combined with MDASH and GPT-5.4 outperformed several competing models in CyberGym vulnerability-discovery testing.
- Microsoft says the system can handle up to 90% of tasks with the smaller model and save 50% in costs compared with its current MDASH offering.
- The model will be available through Project Perception, which Microsoft says can simulate attacks, detect and investigate threats, and help fix vulnerabilities; public preview is scheduled for August 3.
Article Details
- Event Type
- Microsoft announced MAI-Cyber-1-Flash, its first cybersecurity AI model, and said it will be offered through Project Perception.
- Impact
- Microsoft says MAI-Cyber-1-Flash, used with MDASH and GPT-5.4, performed better at vulnerability discovery in CyberGym testing than the named competing models. The company also claims the combination reduces costs by 50% compared with its current best MDASH offering. Project Perception is intended to help organizations simulate attacks, detect and investigate threats, and fix vulnerabilities; no specific vulnerabilities or security incidents are reported.
Vendors
Anthropicwith MDASH and GPT-5.4) topped Google’s recently launched 3.5 Flash Cyber, OpenAI’s GPT-5.6 Sol, and Anthropic’s Mythos 5 in vulnerability discovery.Googleevaluation framework showed that MAI-Cyber-1-Flash (in combination with MDASH and GPT-5.4) topped Google’s recently launched 3.5 Flash Cyber, OpenAI’s GPT-5.6 Sol, and Anthropic’s Mythos 5 in vulnerabilityMicrosoftMicrosoft has unveiled its first cybersecurity AI model, MAI-Cyber-1-Flash, which the company claims got significantly better results in finding vulnerabilities than its main competitors.OpenAI(in combination with MDASH and GPT-5.4) topped Google’s recently launched 3.5 Flash Cyber, OpenAI’s GPT-5.6 Sol, and Anthropic’s Mythos 5 in vulnerability discovery.
Products
3.5 Flash Cyberthat MAI-Cyber-1-Flash (in combination with MDASH and GPT-5.4) topped Google’s recently launched 3.5 Flash Cyber, OpenAI’s GPT-5.6 Sol, and Anthropic’s Mythos 5 in vulnerability discovery.GPT-5.4-CyberCyberGym cybersecurity evaluation framework showed that MAI-Cyber-1-Flash (in combination with MDASH and GPT-5.4) topped Google’s recently launched 3.5 Flash Cyber, OpenAI’s GPT-5.6 Sol, and Anthropic’s Mythos 5 inGPT-5.6 Sol(in combination with MDASH and GPT-5.4) topped Google’s recently launched 3.5 Flash Cyber, OpenAI’s GPT-5.6 Sol, and Anthropic’s Mythos 5 in vulnerability discovery.MAI-Cyber-1-FlashMicrosoft has unveiled its first cybersecurity AI model, MAI-Cyber-1-Flash, which the company claims got significantly better results in finding vulnerabilities than its main competitors.Mythos 5and GPT-5.4) topped Google’s recently launched 3.5 Flash Cyber, OpenAI’s GPT-5.6 Sol, and Anthropic’s Mythos 5 in vulnerability discovery.Project PerceptionMAI-Cyber-1-Flash is being offered through Project Perception, an agentic security offering that enables organizations to simulate attacks, detect and investigate threats, and fix vulnerabilities. Project Perception
Tools
CyberGymAccording to Microsoft, testing in the CyberGym cybersecurity evaluation framework showed that MAI-Cyber-1-Flash (in combination with MDASH and GPT-5.4) topped Google’s recently launched 3.5 Flash Cyber, OpenAI’sMDASHMAI-Cyber-1-Flash, designed to identify challenging vulnerabilities in complex code, has been integrated into Microsoft’s MDASH multi-agent vulnerability identification and remediation harness.