Anthropic noise rattles cyber stocks
Social traders flagged that recent Anthropic announcements — including Claude Code and a Mythos leak — coincided with sharp drops in cybersecurity stocks such as CrowdStrike and Zscaler. Observers also reported central-bank and institutional attention to Mythos’ security implications, with the Bank of Canada among institutions discussing potential risks. (x.com) (x.com)
Shares of major cybersecurity firms fell after Anthropic’s recent disclosures and two separate leaks tied to its Claude products rattled investors. (cnbc.com) On March 27, CrowdStrike Holdings Inc. shares dropped roughly 7 percent and Zscaler Inc. shares slipped about 4.5 percent amid the initial Mythos reporting; Palo Alto Networks and SentinelOne also posted mid-single-digit losses that day. (techbezz.com) (technobezz.com) Fortune reported on March 26 that an unsecured content‑management system left nearly 3,000 unpublished Anthropic assets publicly accessible, and those files included a draft post describing an unreleased model called “Claude Mythos.” (fortune.com via tech.yahoo.com) (tech.yahoo.com) Days later, on March 31, Anthropic accidentally published Claude Code version 2.1.88 to the npm registry with a source‑map that exposed about 512,000 lines of TypeScript and a 59.8 MB archive of its CLI source, a lapse discovered and amplified by security researchers. (nodesource.com / dev.to) (nodesource.com) The leaked Mythos draft warned the model could identify “high‑severity vulnerabilities” faster than defenders, a description that prompted regulators and finance firms to flag new systemic cyber risks. (csoonline.com) International and national authorities moved quickly: the Bank of Canada convened major lenders on April 10 to discuss Anthropic’s model and its implications for critical infrastructure. (bloomberg.com) In the United States, Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell summoned bank chiefs to a Washington meeting the week of April 7 to warn about potential cyber threats tied to Anthropic’s Mythos capabilities. (bloomberg.com / cnbc.com) (bloomberg.com) Anthropic has said it is limiting access to Mythos, rolling the model out to a small set of enterprise partners under an initiative called Project Glasswing and describing Mythos as a “step change” in capability. (forbes.com / venturebeat.com) (forbes.com) Market moves and regulator briefings continued into April as banks and cybersecurity vendors assessed the disclosure and Anthropic’s dual operational lapses; follow‑up meetings and internal risk reviews were reported by multiple outlets this week. (reuters via usnews.com / yahoo.com) (money.usnews.com)
