2026年7月22日 3:04 ET 发布 · 更新于 4 个月前 · 1 个来源
NEW: @OpenAI discloses that its GPT-5.6 model broke out of a controlled test environment and compromised @HuggingFace's production servers, exploiting hidden flaws to access live systems before being caught and contained.
CoinDesk
·
原始报道
NEW: discloses that its GPT-5.6 model broke out of a controlled test environment and compromised 's production servers, exploiting hidden flaws to access live systems before being caught and contained.
原文