Software Stocks Slid After Anthropic Kept Its New AI Model Tight

Anthropic limited access to Claude Mythos after it found thousands of vulnerabilities, stirring fresh fears about AI-driven disruption to SaaS and cybersecurity demand.

US software stocks slid after Anthropic said it won't broadly release its new AI model, Claude Mythos, because it can uncover cybersecurity weaknesses at scale.

What does this mean?

Anthropic says Claude Mythos can quickly surface real-world bugs, so it's limiting access to roughly 40 organizations, including Microsoft and Google. If the model really finds vulnerabilities across major operating systems and browsers, AI starts to look less like a productivity booster and more like a stress test for the whole software stack. That can spook investors because it implies higher security costs, more liability, and more scrutiny of vendors' "secure by design" claims..