Anthropic's Triple Shock: Mythos Too Dangerous, Revenue Surpasses OpenAI, Software Stocks Crash

Anthropic made major AI-related announcements this week, including refusing to publicly release its powerful Mythos model due to cybersecurity risks, surpassing OpenAI in revenue, and causing a software stock market crash.

Details

Anthropic made a series of major announcements this week that have significantly impacted the AI industry. First, the company officially released its powerful Mythos model through Project Glasswing, but refused to make it publicly available. This unprecedented move was due to Mythos autonomously discovering thousands of previously unknown zero-day vulnerabilities across major operating systems and web browsers. Anthropic determined that a public release would 'significantly amplify cybersecurity risks.' Instead, the company granted limited whitelist access to 11 organizations, including tech giants like AWS, Google, and Microsoft. Anthropic also committed $100M in Mythos usage credits and $4M to open-source security organizations. This decision has sparked debates around who should decide which AI models are 'too dangerous' to release. Meanwhile, Anthropic's annual recurring revenue has soared from $9B at the end of 2025 to over $30B, officially surpassing OpenAI as the highest-revenue AI company. The company also signed a massive 3.5GW TPU compute contract with Google and Broadcom, positioning itself for the AI compute arms race. These developments led to a 2.6% single-day drop in the S&P 500 Software & Services Index, as investors feared the implications of an AI model that can rapidly discover vulnerabilities in legacy software. This has accelerated the shift towards AI-native security tools as a new investment thesis.