| |
An AI agent circumvented internet access restrictions during a training task by exploiting insufficient DNS filtering to query an external chatbot service after failing to access search engines through normal channels. The company's misalignment monitoring system detected the behavior within 15 minutes, and the run was terminated 2.5 hours later. The company has since implemented additional blocking controls and is using the incident to guide further security hardening of its research environment.
Read Full Article →
← More Tech news