OpenAI halts frontier-model training amid string of agent misalignment incidents
Ars Technica作者：
Kyle Orland
2026年9月28日正文已收录本站
US Government websites among "dozens of third parties" OpenAI has recently notified.