Cases · Did an AI agent act: yes · Kind of harm: third party security · Model: Claude

What happened	What went wrong	Severity	Earliest source
Anthropic Claude models gained unauthorized access to real third-party systems during evaluations Anthropicmoderate · earliest source 2026-10-22	security breach	moderate	2026-10-22
Anthropic internal evaluation agents exploit internet-facing websites, including U.S. government sites Anthropicmoderate · earliest source 2026-10-10	AI pursued the wrong goal	moderate	2026-10-10
AI agent (ARTEX with Claude) used in cyberattacks targeting South Korean banks Anthropicmajor · earliest source 2026-10-05	security breach	major	2026-10-05
Claude Code used by Yemen-based actors to develop guided-weapon GNC software Anthropicmajor · earliest source 2026-09-14	produced harmful content	major	2026-09-14
Claude in Chrome indirect prompt injection chain: inbox exfiltration, Drive persistence and account takeover Anthropicminor · earliest source 2026-08-05	security breach	minor	2026-08-05
Jailbroken Claude used to breach Mexican government networks and steal taxpayer and voter records Anthropicmajor · earliest source 2026-03-02	security breach	major	2026-03-02
Nx s1ngularity: malicious npm packages drove local AI coding CLIs to enumerate secrets for exfiltrationmajor · earliest source 2025-08-27	security breach	major	2025-08-27