Anthropic's Leak: Cybersecurity Selloff and IPO Liquidity Flow

The leak triggered an immediate and sharp selloff in cybersecurity stocks on March 27. CrowdStrike shares fell 7%, while Palo Alto Networks dropped 6%. Other major names like Zscaler and Okta also saw losses of over 3%. This direct price action was a market reaction to the leaked details of Anthropic's new model, Claude Mythos.

The leaked documents, stored in an unsecured and publicly-searchable data cache, exposed nearly 3,000 internal assets. The draft blog posts described the new model as "by far the most powerful AI model we've ever developed" and highlighted its advanced capabilities in cybersecurity tasks. The core concern for investors was that a model this powerful, capable of automated discovery of zero-day vulnerabilities, could accelerate cyberattacks faster than defenders can respond.

This created a clear liquidity flow: money moved out of cybersecurity equities as the perceived defensive moat of the sector appeared to erode. The market's verdict was swift, with the selloff directly tied to the leaked risk profile of a new AI tool that could undermine the very businesses it was meant to protect.

Anthropic's Capital Flow and Valuation

Anthropic is preparing for a massive liquidity event, with executives discussing an initial public offering by Q4 2026. The planned offering is expected to raise more than $60 billion, positioning it as one of the largest IPOs in history. This capital requirement underscores the extreme financial demands of the AI race, where companies like Anthropic and OpenAI are spending tens of billions to develop and run their models.

The company's recent fundraising reflects this valuation scale. In February, Anthropic closed a $30 billion funding round at a $380 billion valuation. This was followed by a much larger round for its rival OpenAI, which raised $120 billion at an $850 billion valuation earlier this month. The leak confirmed Anthropic is actively testing its most powerful model to date, Claude Mythos, which internal documents describe as a "step change" in performance.

The irony is stark. A company building an AI with unprecedented cybersecurity risks exposed its own most advanced model through a basic human error. This incident highlights the dual-use danger of such technology, where tools designed for defense can be weaponized. For investors, the leak adds a layer of operational risk to an already immense capital story.

Liquidity Catalysts and Market Risks

The primary catalyst for Anthropic's thesis is its planned initial public offering by Q4 2026. This event will force a public valuation of its model capabilities against its security risks, with the company targeting a raise of more than $60 billion. The IPO's scale will be a direct test of market appetite for a company whose most powerful product, as revealed by the leak, poses a fundamental threat to a major defense sector.

The major near-term risk is heightened regulatory scrutiny. The leak's revelations about a model capable of exploiting software vulnerabilities faster than defenders can respond may trigger calls for safeguards on powerful AI. This could lead to new rules that slow Anthropic's rollout or increase its compliance costs, adding a layer of uncertainty to its growth trajectory.

For the selloff in cybersecurity stocks, the path forward hinges on market perception. If the sector is viewed as a net negative for long-term revenue due to AI-driven offensive capabilities, the pressure may persist. The market's verdict will be shaped by Anthropic's own security-focused rollout plan and whether it shares test results to help defenders, as promised in the leaked drafts.