AI BriefingMAI-Cyber 1
Key point
Microsoft has announced MDASH, a multi-agent system for identifying and resolving security vulnerabilities, along with a new model, MAI-Cyber-1-Flash.
Details
Microsoft has unveiled MDASH, a multi-agent vulnerability identification and remediation harness for identifying and resolving security vulnerabilities, and the MAI-Cyber-1-Flash model that operates within it.
MAI-Cyber-1-Flash is a code-centric, security-specialized model designed to find challenging vulnerabilities in complex codebases. This model outperformed Mythos, Gemini, GPT, and others on the CyberGym benchmark, and in particular, when combined with GPT-5.4 within the MDASH system, it delivers a 50% cost reduction compared to existing models.
Key features include the following:
- Efficient model tiering structure: MAI-Cyber-1-Flash efficiently handles about 90% of all tasks, while high-cost models like GPT-5.4 are allocated only to the remaining 10% of high-difficulty tasks, striking an optimal balance between performance and cost.
- Perception launch: Microsoft also launched Perception, an agent-based security system that helps security professionals continuously monitor, patch, and block threat vectors.
- Data and harness optimization: The model achieves unrivaled performance by leveraging Microsoft's decades of accumulated security signals (Identity, Endpoint, Cloud, Network) and real-world exploit/remediation data for training.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Australian PM Reports OpenAI Agent Hacked Medicare Portal in Believed First Known AI-Driven Government Breach