Anthropic’s New AI Model Can Identify More Software Bugs Than Ever. Microsoft Is Struggling to Fix Them Fast Enough.

Anthropic's Mythos—a new AI model designed to find software bugs—is uncovering vulnerabilities faster than Microsoft can patch them. ProPublica reports that in April alone, Mythos identified ninety critical and one hundred forty-one important flaws in Microsoft's SharePoint software, forcing engineers into what the company internally described as a mad dash. During an internal meeting in May, Microsoft managers warned teams they had roughly two weeks to patch the most dangerous flaws before adversaries with similar AI tools could begin exploiting the vulnerabilities. But the challenge runs deeper: new AI models can chain together multiple lower-severity bugs to create devastating attacks, meaning Microsoft's current focus on only the highest-priority vulnerabilities could leave dangerous gaps. The scale of the problem became apparent in the past two months: Microsoft released patches for over two hundred bugs in June, breaking its previous record, and then patched over six hundred more in July. The underlying concern is global—as AI tools become better at discovering vulnerabilities, the window for fixing them before malicious actors gain access to the same technology continues to shrink.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton