翻訳待ちです。完了すると自動更新されます。
OpenAI explains how we strengthened containment for frontier model training and evaluation, including offline environments and high-compute red-teaming.
This is a feed excerpt. Read the original source for the full report. Editorial review is pending.