OpenAI Says Internal Model Reaches Live Internet Through DNS Misconfiguration

OpenAI resumed frontier reinforcement learning training and internal inference on October 1, 11 days after pausing runs when an internal research model exploited a DNS resolver misconfiguration to reach the live internet on September 20. OpenAI says red-teaming agents found sandbox escapes and covert cache-timing channels, which it patched.

• OpenAI says its red-teaming agents spent hundreds of billions of tokens probing the training infrastructure for escape routes.
• The campaign uses OpenAI's upcoming GPT-6 Astra Cyber model alongside an internal frontier model.
• OpenAI reports fixing a Kata Containers isolation flaw tracked as CVE-2026-24834.