AI Daily: Cybersecurity stocks sink after Anthropic model leak

Catch up on the top artificial intelligence news and commentary by Wall Street analysts on publicly traded companies in the space with this daily recap compiled by The Fly.

NEW ANTHROPIC MODEL: Cybersecurity stocks are moving lower after Fortune reported that Anthropic inadvertently revealed details of an upcoming model release "in what appears to be a significant security lapse." A draft blog from Anthropic showed it is testing a new AI model that has powerful cyber capabilities, Beatrice Nolan of Fortune reports. A spokesperson from Anthropic confirmed to the website that it is training and testing a new model. Shares of CrowdStrike (CWRD), Palo Alto Networks (PANW), Cloudflare (NET), Zscaler (ZS), and Okta (OKTA) traded lower following the news.

Stephens attributed this morning's broad weakness in cybersecurity stocks to news around leaked information regarding a new Anthropic foundation model called "Mythos" and headlines referencing more powerful cyber capabilities. However, based on the firm's review of the reporting and a leaked Anthropic blog, it believes the market is "misinterpreting the news" and contends that the "cyber capabilities" being described relate to the model's potential use in launching cyberattacks, not replacing cybersecurity solutions. The firm, which thinks this leak reinforces the view that AI is "a cyber threat accelerant" that increases the importance of cybersecurity, views today's weakness as a buying opportunity for the sector.

Wells Fargo also viewed the selloff in cybersecurity stocks on Anthropic's leaked Claude Mythos details "as another clear misfire by the market." The firm recommended buying network security names Palo Alto Networks and Zscaler, calling them "large AI beneficiaries with minimal disruption risk." Anthropic highlighted the cyber risk Mythos and upcoming models pose, which "actually underscores the AI opportunity for vendors," Wells told investors in a research note. The firm also views CrowdStrike as well positioned.

GOING PUBLIC: Anthropic executives have discussed an IPO as soon as Q4, and bankers vying to take the company public expect it to raise more than $60B, The Information's Valida Pau and Sri Muppidi report.

DOD SUIT: A U.S. judge has granted Anthropic's request for a preliminary injunction in its suit against the Trump administration over the DOD's decision to blacklist the company, CNBC's Ashley Capoot and Jeffrey Kopp report. Anthropic sued the administration to try to reverse the Defense Department's decision to blacklist the company after contract talks fell apart. Judge Rita Lin questioned the U.S. government about the "attempt to cripple" the AI startup in a hearing on Tuesday.

REVENUE FROM ADS: OpenAI has surpassed $100M in annualized revenue from ChatGPT ads, has expanded to more than 600 advertisers, and plans to launch self-serve advertiser access in April, The Information's Stephanie Palazzolo reports.

REALITY LABS: Meta's (META) Reality Labs is reorganizing employees into AI-native "pods" focused on specific outcomes, as part of a shift toward a flatter organization, Business Insider's Charles Rollet reports. A leaked memo describes an overhaul of roles, titles, and team structures across a 1,000-employee team within Meta's Reality Labs, adding that it is part of a broader, aggressive push by Meta to adopt small teams and use AI.