Back to feedArs Technica - All contentTechnologynewsOpenAI halts frontier-model training amid string of agent misalignment incidentsMonday, September 28, 2026 Kyle Orland View original––– US Government websites among "dozens of third parties" OpenAI has recently notified.
