Skip to main content

Share story

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
AI Science
AI Science

ALTO recipe narrows four-bit model training gap in simulated tests

Researchers report in a preprint that ALTO, an open-source training recipe for the 20-billion-parameter GPT-OSS-20B model, reduced the extra training steps needed with four-bit arithmetic. On the MLPerf Small MoE Training benchmar...

AI Policy
AI Policy

UK commits £100m to AI biosecurity lab

The UK government will put £100m into a new AI Biosecurity Lab to detect biological threats and accelerate vaccine and treatment research, the Cabinet Office said Saturday. OpenAI and Anthropic will contribute up to £200m between ...

AI Capital
AI Capital

OpenAI lowers funding valuation target to $1.25tn

OpenAI has lowered the valuation it seeks in a new funding round to $1.25tn from $1.4tn after some investors resisted the higher price, the Financial Times reported, citing documents it reviewed. The company is close to raising u...

AI Policy
AI Policy

UK commits £100 million to planned AI biosecurity laboratory

The UK government has announced £100 million to build a laboratory that will use AI to detect and monitor biological threats, ChainCatcher reported, citing Bloomberg. OpenAI and Anthropic will provide up to £200 million in AI mode...