U.S. software stocks slid sharply on Thursday after Anthropic said it would not proceed with a wide release of its newest AI model, Claude Mythos, citing concerns the model could reveal hidden cybersecurity weaknesses. Anthropic said it will grant access to roughly 40 companies, including Microsoft and Google, rather than opening the model to the public because it has already identified thousands of vulnerabilities, including some affecting every major operating system and web browser.
Market participants interpreted the announcement as a renewed warning about the vulnerability of existing software and the pace of AI advances versus legacy vendors. "If Mythos is that strong and that powerful and it’s exposing these vulnerabilities that have been around for years, it just shows one, the weakness of the current software that’s out there and two, that AI is still making incredible progress versus legacy software companies," said Michael O’Rourke, chief market strategist at JonesTrading.
The S&P 500 Software and Services Index has dropped nearly 26% year-to-date, and it fell another 3.1% on Thursday as investors weighed the implications for software-as-a-service providers that sell subscription-based products.
Cybersecurity stocks were among the notable decliners in morning trade. Cloudflare, Okta, CrowdStrike and SentinelOne each fell in a band between 4.7% and 7.7%. Zscaler was one of the S&P 500’s largest losers, down 8.6% after brokerage BTIG downgraded the stock to neutral from buy, citing worries over demand and the potential for increased competition.
"We’re getting back to being concerned about the prior software-specific concerns stemming from AI and private credit that are coming back to the fore," said Steve Sosnick, chief market analyst at Interactive Brokers.
Several enterprise software firms also gave back ground on Thursday. Atlassian, Workday, Adobe, Salesforce and Intuit each slid within a range of 3.7% to 6.8% as investors reacted to the broader market reassessment of AI-related risk to incumbents.
Market context and mechanics
The immediate market reaction concentrated on software and cybersecurity names, highlighting investor sensitivity to developments that could accelerate competitive disruption or surface latent risk in widely used platforms and applications. The selective access granted by Anthropic - to about 40 firms including major technology companies - underscores the company’s caution in rolling out a model it says has uncovered extensive vulnerabilities.
What to watch next
- Whether broader access to AI models is expanded or remains restricted based on further vulnerability findings.
- Earnings and demand signals from SaaS companies that could indicate whether AI-driven disruption is already affecting renewal and new-business trends.
- Responses from cybersecurity vendors and incumbent software firms as they assess and disclose any newly identified risks.