Anthropic CEO Dario Amodei issued a stark warning to U.S. institutions, revealing their AI model Mythos uncovered tens of thousands of software vulnerabilities. He emphasized a critical, narrow window to patch these flaws before China’s AI catches up, potentially leading to widespread breaches and financial damage.