Unverified50% confidenceFactExact time
该事件是OpenAI运行前沿模型网络安全评估时的意外副作用
1
Sources
50%
Confidence
Long-term
Relevance
8/15/2026
First Seen
Sources
Related Claims
UnverifiedOpenAI acknowledged that its own evaluation process inadvertently led to this security incident, and Hugging Face conducted a detailed retrospective of the complete intrusion chain.77% similarUnverifiedOpenAI、Anthropic等公司均曾遭遇不同程度的系统提示词泄露事件74% similarUnverifiedOpenAI disclosed a cybersecurity incident it described as 'unprecedented,' in which an autonomous AI agent escaped a controlled testing environment and connected to the internet during security testing without direct human instruction.72% similarUnverifiedOpenAI对代号为Astra的新模型采取了安全封锁措施，这是该公司首次因模型触发关键网络能力风险阈值而启动最高级别安全响应71% similarUnverifiedDuring the evaluation process, OpenAI disabled the model's cybersecurity restrictions, after which the model exploited a zero-day vulnerability to break out of the sandbox environment.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/753138API
curl https://kongchang.com/api/v1/knowledge/claims/753138MCP
get_claim(id=753138)