Cybersecurity Stocks Tumble After Anthropic's Claude Mythos AI Leak Sparks Market Fears

Major cybersecurity stocks took a sharp hit Friday following reports that Anthropic accidentally exposed details about its next-generation AI model before its official launch. CrowdStrike fell 7%, Palo Alto Networks dropped 6%, Zscaler slid 4.5%, while Okta, SentinelOne, and Fortinet each lost around 3%.

The selloff was triggered by a Fortune report revealing that internal documents about Anthropic's unreleased model, codenamed "Claude Mythos," were mistakenly stored in a publicly accessible and searchable data cache. Anthropic attributed the exposure to a human configuration error within its content management system. The leaked material introduced a new model tier called "Capybara," designed to surpass the company's existing top-tier Opus model in coding, academic reasoning, and cybersecurity performance. Anthropic described Claude Mythos as the most powerful AI system the company has ever built.

An Anthropic spokesperson confirmed the company is actively developing a model with significant advances in reasoning, coding, and cybersecurity, emphasizing a careful and deliberate release strategy. The leaked draft reportedly warned that the model's cyber capabilities significantly outpace any existing AI, raising concerns that it could trigger a wave of advanced exploits that overwhelm traditional defense systems.

Wall Street analysts quickly weighed in on the broader implications. Raymond James' Adam Tindle warned that AI-driven attack capabilities could erode conventional cybersecurity advantages, increase attack complexity, and shift how organizations allocate security spending. Evercore ISI's Kirk Materne cautioned that the sector faces a prolonged adjustment period as long as it continues reacting sharply to every AI development. Stifel's Adam Borg framed Claude Mythos as potentially transformative, suggesting it could elevate even low-level hackers to near nation-state threat levels.

Despite near-term market anxiety, analysts broadly agree that escalating AI-powered threats will push cybersecurity higher on enterprise IT agendas, ultimately accelerating investment in modern, AI-integrated defense solutions away from outdated legacy tools.