technology
OpenAI details expanded monitoring for frontier AI agents
OpenAI says it has expanded monitoring of its most capable AI agents across training, evaluation and internal use, with modified versions for some external deployments. The system combines automated checks, investigations and human response to flag potentially misaligned behaviour.