Microsoft (MSFT) Launches AI-Powered MDASH Scanner to Counter Evolving Cyber Threats

Microsoft Corp. (MSFT) launched an expanded preview of its Microsoft Security multi-modal agentic scanning harness, or MDASH, on June 2, 2026, to defend against AI-enabled cyberattacks. The platform deploys AI agents to rapidly identify software vulnerabilities, with Vasu Jakkal, Microsoft’s CVP of security, warning that “the time between vulnerability discovery to exploitation essentially collapses” amid the rise of adversarial AI. MDASH uses more than 100 configurable models to balance cutting-edge performance and cost efficiency. The move comes as Anthropic’s Claude Mythos Preview model has demonstrated alarming proficiency at finding flaws in all major browsers and operating systems, despite not being developed for cybersecurity. Microsoft was among 40 companies that gained early access to Mythos under a partnership; Anthropic expanded that to 140 additional firms on the same day. OpenAI has also developed a cybersecurity-focused model, GPT-5.4 Cyber, signaling an industry-wide AI arms race.