Skip to main content
AI

MAI-Cyber-1-Flash inside MDASH

MAI-Cyber-1-Flash inside MDASH Image: Primary
Microsoft announced MAI-Cyber-1-Flash inside MDASH, its multi-agent vulnerability identification and remediation harness, saying the combined system delivers world-class performance at 50% of the cost of leading models. The company said MAI-Cyber-1-Flash was built to find challenging vulnerabilities in complex codebases and has been deeply integrated into MDASH. Microsoft said the unified system delivers 96% on CyberGym, a benchmark for evaluating how systems reason over large codebases to find real vulnerabilities, which is 12 percentage points above Mythos. The system was designed to efficiently handle up to 90% of all tasks, enabling MDASH to use larger models such as GPT-5.4 for the 10% of exceptionally hard tasks. Microsoft also launched Perception, an agentic security system that provides teams of agents for a variety of security workflows in MDASH to continuously monitor, patch, and close new threat vectors. The company said Perception will soon use MAI-Cyber-1-Flash for many more security workflows beyond software vulnerability work. Microsoft said the model was developed with a security-first calibration, rigorously evaluated by Microsoft's AI Red Team, and independently assessed by a third party.
Sources
In this story
Published by Tech & Business, a media brand covering technology and business. This story was sourced from microsoft.ai and reviewed by the T&B editorial agent team.
Back to Newswire
Keep reading
Full wire
Infrastructure Products
Infrastructure Products

Blackstone's AirTrunk in talks for S$2B loan to fund Singapore REIT listing

AirTrunk, the data center operator owned by Blackstone Inc., is in talks with banks to borrow about S$2 billion (US$1.6 billion) to help finance the public listing of a real estate investment trust backed by its assets, according ...

Policy Security
Policy Security

EU Cyber Resilience Act vulnerability reporting obligations take effect

The EU Cyber Resilience Act's vulnerability reporting obligations to ENISA arrive on September 11, according to an OpenSSF tech talk recap. The full regulation lands in December 2027. OpenSSF cited a 2026 readiness report finding...

AI Infrastructure
AI Infrastructure

AWS makes SageMaker HyperPod model caching generally available

Amazon Web Services said model caching for SageMaker Inference on HyperPod is now generally available in all regions where HyperPod is offered. The feature pre-loads model weights and inference-server container images onto cluste...