An Anthropic AI model autonomously submitted a fabricated homicide tip to the Philadelphia Police Department during a routine test, and the company waited two months before notifying authorities. The incident is one of several cases where Anthropic's AI agents took unintended real-world actions, including submitting live government forms and exploiting university servers. The White House's newly formed Super Intelligence Force is now demanding immediate incident reporting from AI companies. https://qz.com/anthropic-ai-model-fake-murder-tip-philadelphia-police-101026

OpenAI took about 2.5 hours to stop an AI agent that had managed to break out of the training environment and into the public Internet. The company’s monitoring systems detected the issue just a few minutes after it occurred. The corporation detailed this September 20 incident in a report published recently. This incident gained widespread […] Post 2.5 Hours of Chaos: OpenAI Shut Down Its Most Powerful AI Models After a New Incident at Root-Nation.com.

Google disclosed an incident in which its Gemini AI, during testing, escaped its isolated environment and gained access to the computer systems of three third-party companies. The model independently generated passwords and used an online repository of publicly available login credentials. The incident occurred in May 2026 during a capture-the-flag test conducted by the AI […] Post Gemini Got Out of Control: Google’s AI Guessed Passwords and Hacked Into the Systems of Three Companies at Root-Nation.com.

A security engineer’s analysis of the OpenAI-Hugging Face incident, what failed in the architecture and incident response, and the lessons security teams can learn from it.

https://www.osintinvestigate.com Hundreds of OpenAI AI agents reportedly coordinated through an unauthorized channel and attacked Hugging Face during a cybersecurity evaluation. We examine reward hacking, AI coordination, attempts to manipulate records, and what this incident means for AI safety and cybersecurity.

YouTube video by B.C. Begley

Hugging Face reported a breach from an autonomous AI agent. Experts discuss the incident response effectiveness and implications for future security.