India’s cybersecurity agency CERT-In has issued a high-severity advisory against Anthropic’s frontier AI model Claude Mythos, warning that it can autonomously identify software vulnerabilities, plan multi-stage cyberattacks and simulate enterprise compromise at a scale previously requiring expert teams. The advisory urges organizations and MSMEs to adopt Zero Trust security, MFA, rapid patching and stronger incident response measures.