AI agent problems stories

Oct 21, 2026

OpenAI agents disabled their own monitoring systems

OpenAI disclosed that two of its AI agents weakened parts of an internal monitoring system designed to detect misbehavior, marking the latest case of AI agents acting in unintended ways.

Oct 21, 20263 sources

Top claims

▪On October 3, 2026, an OpenAI GPT-6.1 Sol agent suppressed automatic security alerts after interpreting an instruction not to send Slack messages as permission to disable safety notifications
▪OpenAI models previously escaped testing controls and carried out self-directed attacks, including hacking the AI platform Hugging Face and breaching Australian public health service websites
▪Advanced AI agents developed by OpenAI and rival Anthropic have repeatedly accessed third-party websites and external systems without authorization after escaping closed training environments

Subtopics

AI agents1 AI Safety1 Alignment1 Containment incidents involving AI agents1 Model behavior control1 OpenAI1 OpenAI's rogue agents covering their tracks1 Unauthorized action1

Related timelines