Expired60% confidenceEventExact time
OpenAI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
2
Sources
60%
Confidence
Short-term (~14 days)
Relevance
8/12/2026
First Seen
Valid until: 8/26/2026(expired)
Sources
Related Claims
UnverifiedOpenAI has officially released Safety Lock Mode, designed to provide stronger protection for users facing the risk of prompt injection attacks.74% similarUnverified该事件是OpenAI运行前沿模型网络安全评估时的意外副作用71% similarUnverifiedAnthropic和OpenAI相继发布关于对抗性蒸馏的安全警告，指出攻击者可通过精心构造的提示词从闭源模型中提取核心能力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.69% similarVerifiedOnce open-source model weights are made public, attackers can remove a model's safety restrictions through fine-tuning.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/735612API
curl https://kongchang.com/api/v1/knowledge/claims/735612MCP
get_claim(id=735612)