OpenAI says it has strengthened isolation and monitoring for frontier-model research, with reinforcement-learning runs using an offline environment.
In its October 23 report, the company describes pausing its most capable training and evaluation workloads after a September 20 incident, then resuming frontier reinforcement learning and internal inference on October 1 after additional defenses and testing. High-compute adversarial testing is now part of its safety cases. These findings concern internal research environments; the report does not establish that released customer products experienced the same incidents.
