What is Anthropic Claude Mythos? Everything to know about viral leaked AI model that set alarms in cybersecurity

In an official statement, Anthropic confirmed that the exposure was unintentional and attributed it to a misconfiguration rather than a sophisticated breach.

Cybersecurity stocks tumbled sharply on March 27 after Anthropic accidentally leaked details of its unreleased next-generation model, codenamed Claude Mythos. That's a name that everyone should take seriously, even its rivals. The leaked data revealed internal documents describing the model as "by far the most powerful AI model we've ever developed" -- one that promises to address cybersecurity risks on a level that none of today's solutions can, thanks to its advanced capabilities in detecting, exploiting, and responding to threats.

Information pertaining to Claude Mythos leaked when draft blog posts and related assets were stored in a publicly accessible data cache. The leak was narrowed down to a 'human error' in Anthropic's content management system configuration.

The leaked materials also included details of a new AI tier called "Capybara," positioned above the current Claude Opus model, along with information about an exclusive CEO event.

As soon as the leak surfaced, investors were ready, interpreting the model's superior performance on cybersecurity benchmarks as a potential threat to traditional enterprise security providers. The sell-off wiped out approximately $14.5 billion in market value across major players in a single trading session.

Claude Mythos: Making an impact before launch

As soon as details about the model leaked, the market went harsh on cybersecurity firms. CrowdStrike (CRWD) fell 5.85%, erasing roughly $5.5 billion in market cap. Palo Alto Networks (PANW) dropped 6.43%, losing about $7.5 billion. Zscaler (ZS) declined 5.89%, with $1.35 billion wiped out, while Tenable shed 9.70%, or $185 million. The Global X Cybersecurity ETF also slid more than 2.7%.

The leaked documents highlighted Mythos's ability to outperform existing tools in areas such as vulnerability exploitation and automated threat response - capabilities that could reduce reliance on large teams of security professionals and expensive software suites.

Hence, Claude Mythos could make the cybersecurity sector extinct if it delivers on its promises.

Anthropic admits leak, says model remains in testing

In an official statement, Anthropic confirmed that the exposure was unintentional and attributed it to a misconfiguration rather than a sophisticated breach. The company has not yet detailed Mythos's full capabilities or release timeline, even though internal previews (from the leaked documents) described it as a "step-change" in performance, particularly in cybersecurity tasks.

"We're developing a general-purpose model with meaningful advances in reasoning, coding, and cybersecurity. Given the strength of its capabilities, we're being deliberate about how we release it. As is standard practice across the industry, we're working with a small group of early access customers to test the model. We consider this model a step change and the most capable we've built to date," said a spokesperson.

The leak incident marks the second time Anthropic dominates the headlines, triggering activity in the market. Earlier in the year, Anthropic's release of Claude Code had hit the stock market globally, sending tremors throughout the industry. This comes on the heels of Anthropic going on to sue the Pentagon after the company was declared a 'supply chain risk', thereby severing its involvement in defence use cases with the US Department of Defense.

Will Claude Mythos decimate the cybersecurity space?

While the immediate reaction from investors reflects fears of disruption, Anthropic is interested in helping cybersecurity defenders rather than fully replacing them. Enterprise solutions rely on complex integrations, compliance, and human oversight that go beyond raw model performance. An AI model helping here could only strengthen cybersecurity efforts at all levels.

At the same time, Claude Mythos could end up being a potential threat to complex defence systems as it tries to find vulnerabilities in their defences.

For now, Claude Mythos remains a myth in the cybersecurity space and we only have to wait for an official confirmation from Anthropic.