Latest Anthropic horror story chills with tales of kamikaze drone swarms and bioweapons research

Anthropic’s report documents widespread misuse of its Claude AI models from Dec 2025–Aug 2026 by state-aligned and criminal actors: an SVR-linked APT automated entire cyber kill chains against embassies, think tanks, and defense targets; ShinyHunters-scaled supply-chain/data theft (including >2,100 Azure AD tokens and ~200 impacted SaaS customers); and multiple cases of AI-assisted biological-risk research and conventional weapons development (guided rockets, anti-torpedo systems, and kamikaze drone swarms), leading Anthropic to block accounts and share threat intelligence.