AI's Next Shock: Anthropic's New Model Accidentally Leaked, Cybersecurity Stocks Plummet at Market Open

An unreleased AI model from Anthropic has been accidentally exposed through a data leak, reigniting market concerns about AI disrupting the cybersecurity industry. Cybersecurity stocks fell across the board in pre-market trading on Friday.

According to Fortune magazine, Anthropic is developing and has begun testing a new AI model with early users. The company acknowledged that descriptions of the model were leaked after being stored in a publicly accessible data cache.

An Anthropic spokesperson characterized the new model as a "step-change breakthrough" in AI performance, calling it the "most powerful model the company has built to date." Notably, the leaked documents mentioned that the model could pose "unprecedented cybersecurity risks."

Impacted by this news, the cybersecurity sector experienced a rapid decline at the market open. CrowdStrike fell 6.0%, Zscaler dropped 4.7%, Palo Alto Networks fell 4.6%, Okta dropped 4.26%, Fortinet fell 4.0%, and Cloudflare fell 3.9%. The Global X Cybersecurity ETF, which tracks the entire sector, declined 2.7%.

The Leak: Draft Document Exposed in Public Database

According to Fortune, a draft blog post detailing the new model was stored in an unencrypted, publicly searchable data repository until Thursday evening.

The document indicated that the new model is named "Claude Mythos," and Anthropic believed its cybersecurity risks exceeded those of any previous model.

Anthropic confirmed the leak and the existence of the new model, stating that it is currently in the testing phase with early access customers and has not been officially released. A company spokesperson emphasized that this new model represents a significant leap in the company's AI capabilities.

Sector Logic: Leap in AI Capabilities Directly Impacts Traditional Security Vendors

The decline in cybersecurity stocks reflects the market's long-standing structural concern about AI disrupting the traditional security software industry.

Investors are widely worried that as AI companies like Anthropic and OpenAI release increasingly powerful models, enterprise customers' demand for traditional cybersecurity products may shrink, thereby suppressing these vendors' revenue growth, profit margins, and pricing power.

The mention of "unprecedented cybersecurity risks" in the leaked document has led to a dual interpretation in the market:

On one hand, powerful AI tools could become novel attack vectors, challenging existing defense systems. On the other hand, AI-native security capabilities might partially replace traditional security products, putting pressure on the market share of existing vendors. This logic, combined with a low pre-market liquidity environment, exacerbated the concentrated sell-off in the sector.

Market Background: AI Disruption Trades Make a Comeback

The cybersecurity sector, along with broader software stocks, has been continuously suppressed by the "AI replacement" narrative recently. The exposure of Anthropic's new model has brought this market theme back into the spotlight.

For investors, the core question is: as the security and attack capabilities of AI models advance in tandem, how deep is the moat for traditional cybersecurity vendors, and to what extent will the irreplaceability of their products be eroded?

Currently, the official release date and complete technical details of Claude Mythos have not been disclosed, nor has Anthropic provided further explanation on the specific impact of the leak. The market will continue to monitor the model's future developments and Anthropic's official statements.