By Nate Wolf
Cybersecurity stocks were getting crushed on Friday. Another piece of news from artificial-intelligence lab Anthropic has investors on edge.
Anthropic is developing a new Claude large-language model called Mythos, according to a report from Fortune. Anthropic confirmed the model's existence to Fortune after the publication discovered a data leak. The company didn't immediately respond to Barron's request for comment.
According to the report, the highest-tier system -- dubbed Capybara -- has more advanced cyber capabilities than any other AI model, presenting potential cybersecurity risks.
Cybersecurity stocks plunged. Palo Alto Networks stock fell by 6% on Friday, CrowdStrike dropped by 5.7%, Zscaler lost 6.2%, and Fortinet declined by 3.4%.
The iShares Cybersecurity & Tech exchange-traded fund, which was down 8.1% this year as of Thursday's close, fell another 3.4% on Friday.
Cybersecurity stocks have been sensitive to Anthropic headlines. Last month, the AI company unveiled Claude Code Security, a tool built into its model that scans codebases for vulnerabilities and suggests software patches.
Today's news suggests that cybersecurity firms are potentially incapable of protecting clients against bad actors using AI. And security companies are playing a game of catch-up with Anthropic's AI abilities.
Anthropic's latest model "presages an upcoming wave of models that can exploit vulnerabilities in ways that far outpace the efforts of defenders," the company said in the leaked documents obtained by Fortune.
One of the files Fortune reported on said that Anthropic plans to release the model early to organizations to give them a "head start in improving the robustness of their codebases against the impending wave of AI-driven exploits."
The worry isn't so much that AI will replace cybersecurity tools; rather, malicious actors will use AI to get through the security systems.
Industry executives have attempted to push back on both worries over the last several months.
"As enterprises start putting more critical functionality in the hands of AI, they will want control of AI agents or of their AI infrastructure, and that requires more security," Palo Alto CEO Nikesh Arora told investors in a conference call last month. "So I think generally, it's a positive trend towards more security adoption."
Write to Nate Wolf at nate.wolf@barrons.com
This content was created by Barron's, which is operated by Dow Jones & Co. Barron's is published independently from Dow Jones Newswires and The Wall Street Journal.
(END) Dow Jones Newswires
March 27, 2026 11:02 ET (15:02 GMT)
Copyright (c) 2026 Dow Jones & Company, Inc.
Comments