Claude Mythos Leak Sparks Market Selloff Over AI Cybersecurity Fears
A leaked Anthropic AI model named Claude Mythos triggers sharp declines in software stocks and bitcoin due to alarming vulnerabilities.
Anthropic faces a major crisis after leaked internal documents reveal a powerful new AI model called Claude Mythos. The company currently tests this system with a small group of early access customers and describes it as the most capable model it builds to date. Details surface after an accidental data exposure makes roughly 3,000 internal assets publicly accessible online.
The leaked materials warn that Claude Mythos poses serious cybersecurity risks because it rapidly identifies and exploits software vulnerabilities. This alarming capability threatens to accelerate a dangerous cyber arms race. The documents suggest Anthropic is developing a new model tier called Capybara to house this advanced technology alongside its existing Opus, Sonnet, and Haiku lines.
Financial markets react harshly to the news, with major cybersecurity stocks taking significant hits. Companies like Palo Alto Networks, Crowdstrike, and Fortinet all drop between four and six percent, while the broader iShares Expanded Tech-Software Sector ETF falls nearly three percent. The overnight leak also contributes to a sharp bitcoin tumble, pushing the cryptocurrency back down to $66,000 after briefly approaching $70,000.