Cases · How the AI agent failed: escaped its sandbox · How the AI was used: agentic with tools

Patterns across all cases

What happened	What went wrong	Severity	Earliest source
Anthropic's Claude agent autonomously accessed Georgia voter data hub and bypassed a paywall Anthropicmoderate · earliest source 2026-10-26	operational mistake	moderate	2026-10-26
OpenAI research model escapes to live internet via DNS misconfiguration during training OpenAImajor · earliest source 2026-10-25	security breach	major	2026-10-25
RL-trained agent on Alibaba AI engine team mined cryptocurrency on training server and opened reverse SSH tunnel to external IP Alibabaminor · earliest source 2026-10-21	AI pursued the wrong goal	minor	2026-10-21
DB-GPT 0.8.0 unauthenticated RCE via sandbox fallback and skill_upload directory traversal (CVE-2026-51862, CVE-2026-51869)major · earliest source 2026-10-08	security breach	major	2026-10-08
AI agent (ARTEX with Claude) used in cyberattacks targeting South Korean banks Anthropicmajor · earliest source 2026-10-05	security breach	major	2026-10-05
OpenAI agents bypass website security controls and republish SEC data OpenAImoderate · earliest source 2026-09-26	security breach	moderate	2026-09-26
OpenAI evaluation agent breached Services Australia Medicare portal and wrote to government database OpenAImajor · earliest source 2026-09-24	security breach	major	2026-09-24
Google Gemini accessed systems of three real companies during a misconfigured third-party cybersecurity evaluation Googlemoderate · earliest source 2026-09-19	security breach	moderate	2026-09-19
AISI cyber evaluation: Mythos 5 agent exposed locally-hosted attack tooling to the public internet via tunnelling services Anthropicminor · earliest source 2026-08-04	AI pursued the wrong goal	minor	2026-08-04
Model writes and runs code on external internet service to reach AISI evaluation infrastructure during impossible cyber evaluationminor · earliest source 2026-07-21	security breach	minor	2026-07-21