Unverified50% confidenceDecision RuleExact time
Anthropic表示如果模型达到ASL-3阈值但相应安全措施尚未就绪，公司将暂停模型部署直至条件满足
1
Sources
50%
Confidence
Long-term
Relevance
8/9/2026
First Seen
Sources
Related Claims
Unverified准备框架规定：任何维度达到危急级别的模型须在安全措施到位前停止部署，达到高级别的模型需经安全委员会审批69% similarUnverifiedAnthropic introduced the Responsible Scaling Policy, committing to pause capability scaling when models reach specific danger thresholds until safety measures catch up.63% similarUnverified准备框架规定只有风险评级在中及以下的模型才允许部署，高级别需要额外缓解措施，关键级别的模型完全不得部署或继续开发59% similarUnverified停止条件设计通常面临性能阈值、收益递减检测、资源硬限制三类权衡，成熟系统往往同时引入多种机制并赋予不同优先级59% similarUnverified应为智能体设置明确的终止条件和迭代上限，当连续N次未能通过测试应升级为人工介入58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/718849API
curl https://kongchang.com/api/v1/knowledge/claims/718849MCP
get_claim(id=718849)