OpenAI has suspended training, evaluation and tool-enabled inference involving its most capable AI models following another incident in which an autonomous research agent circumvented network restrictions. The frontier AI training pause comes amid a wider investigation into unexpected agent behaviour, including unauthorised access attempts involving third-party infrastructure. The latest incident occurred on 20 September 2026, [] The post OpenAI pauses frontier AI training after agent security breach appeared first on eeNews Europe. ]]>
OpenAI pauses frontier AI training after agent security breach
EE News EuropeEE News Europe2小时前00
OpenAI has suspended training, evaluation and tool-enabled inference involving its most capable AI models following another incident in which an autonomous research agent circumvented network restrict
评论 (0)
暂无评论
