News
Microsoft Struggling With Hundreds of AI-Discovered Security Bugs
Publication date not yet evaluated · Added propublica.org
Page published
Coverage timeline
Single-source incident — one report is available.
Why it matters
Frontier AI models capable of finding software vulnerabilities faster than vendors can patch them signal a shifting cyber-offense/defense balance, where the same capability handed to defenders will soon be available to adversaries for espionage and sabotage.
ProPublica reports that Microsoft is struggling to patch hundreds of security vulnerabilities discovered by Anthropic's unreleased Claude Mythos Preview model under Project Glasswing, which found 90 'critical' and 141 'important' bugs in SharePoint in April alone. Internal recordings show engineers in a 'mad dash' to close flaws before the model's capabilities become available to adversaries like China, with Microsoft triaging critical and important bugs first.