How Anthropic’s leaked AI model wiped out $14.5B from cybersecurity stocks in one day

Cybersecurity stocks plunged on March 27, after a report by Fortune revealed that AI giant Anthropic is working and testing a new AI model that the company itself feels could pose “unprecedented cybersecurity risks.” The draft blog post of the company also accidentally exposed via an unsecured database, described the model codenamed Calude Capybara as powerful enough to help hackers bypass the exiting cyber defences. According to a report by Bloomberg, this leaked model of Anthropic wiped out nearly $14.5 billion in cybersecurity stocks. The fallout was immediate:* CrowdStrike, Palo Alto Networks, and Zscaler each dropped more than 5%* Cloudflare fell 3.2%

* The Global X Cybersecurity ETF slid as much as 6.1%, erasing $14.5 billion in market value in a single day

Why investors panicked

As per the Bloomberg report, the leaked draft suggested that Anthropic is also delaying the release of the model and it plans to share test results with cyber firms in order to help them prepare defences. Analysts feel that the AI giant is trying to prevent misuse by hackers, but the mere possibility of such risks rattled the investors. Bernstein’s Peter Weed noted: “We think Anthropic is also trying to limit their product being used by hackers. This is good hygiene and a baseline expectation for their product.”

Past incidents heighten concerns

Earlier this year, hackers reportedly exploited Anthropic’s chatbot to attack Mexican government agencies, stealing sensitive tax and voter data. Anthropic said it investigated, disrupted the activity, and banned the accounts involved.The company has also triggered sell-offs before. Just last month, Anthropic unveiled a tool to scan codebases for vulnerabilities, which sent shares of major cybersecurity firms tumbling.Despite the sell-off, some Wall Street voices urged calm. Stephens analyst Todd Weller said the market was “misinterpreting the news” and called it a buying opportunity. Portfolio manager Joe Tigay added: “If they are a cybersecurity threat, don’t you want the best companies that deal with cybersecurity to be fighting it for you?”Anthropic’s leaked model underscores the double-edged nature of advanced AI: while it can strengthen defenses, it can also empower attackers. With cybersecurity stocks already down more than 20% this year, the incident highlights how sensitive markets are to AI’s disruptive potential.