Incidents | Rogue AI Tracker
rogueaitracker.com · 8,611 words · saved by 1 readers
Incidents scored against the Rogue AI Tracker capability rubric, with notable score increases highlighted under capability impact.
Reported: Sep 10, 2026 Occurred: Mar 16, 2026 Anthropic Claude agents took cloud administration and exfiltrated customer data Anthropic reported that suspected ShinyHunters affiliates used Claude agents across live intrusions, including a supply-chain compromise where agents performed nearly all the work, reached full cloud administration in about three hours, and extracted data belonging to hundreds of downstream organizations. A separate compromise used Claude to help escalate access and bulk-export data from thousands of downstream customers. Capability impact Didn't set a new…
saved by
related reading
- Countering misuse of AI: September 2026 / Anthropicanthropic.com
- Pre-deployment auditing can catch an overt saboteuralignment.anthropic.com
- Rogue AI Trackerrogueaitracker.com
- Disrupting the first reported AI-orchestrated cyber espionage campaign \ Anthropicanthropic.com
- Incident Report: unsanctioned agent behaviour during cyber testing | AISI Workaisi.gov.uk
- Security incident disclosure — July 2026huggingface.co
- Investigating three real-world incidents in our cybersecurity evaluations \ Anthropicanthropic.com
- An alignment assessment of recent cybersecurity incidentsanthropic.com
- Brief independent investigation of agents’ behavior, reasoning and collaboration in the OpenAI / Hugging Face hacking incidentmetr.org
- Discovery of a new OpenAI agent message boardcollusion.wiki
- Black Hat USA 2026 | The 'Breaking' News: The OpenAI–Hugging Face Incidentyoutube.com
- Your AIs don't do what you want. This is really badrewardhacking.org