As AI continues to accelerate cyber risk globally, a new generation of frontier AI models is reshaping the cybersecurity landscape, reducing the time organisations have, to prepare for AI-powered cyberattacks.
In response to this rapidly evolving threat environment, Microsoft has announced a new multi-model, AI-driven vulnerability discovery system embedded with more than 100 individual agents. Codenamed MDASH, this technology provides a harness for defenders to use multiple increasingly powerful models combined with agent reasoning and marks a significant step forward in how organisations identify and respond to security risks in the age of advanced AI.
MDASH was initially developed to leverage multiple third-party AI models to achieve security tasks and protect Microsoft code. It is an agentic security system that is now being made available to others in a private usage programme. This approach reflects a broader industry shift toward AI-driven, collaborative security models at scale.
AI acceleration reshapes the cyber threat landscape
Microsoft has identified five areas where advanced frontier models are disproportionately advantaged compared to traditional defenders – patching, open-source software, custom code created by an organisation, systems publicly exposed and baseline security hygiene.
