SITUATION DETECTED: OpenAI says its largest planned frontier RL run remains on hold while it conducts smaller-scale training and evaluations to assess model behavior, validate safeguards, and gather more evidence of alignment before proceeding.