跳到正文
ENTITY PROFILE / 实体档案

METR

RSS

Anthropic gives employee‑level access to six METR researchers METR will publish its own assessments of Anthropic’s risks, incidents, and practices Anthropic cannot veto METR’s publication of its assessments Anthropic encourages its employees to share information with METR researchers as they would with a fellow employee METR investigated OpenAI Anthropic provides employee‑like access to METR METR investigated the hack of Hugging Face METR investigated rogue OpenAI models METR is funded by the UK’s AI Security Institute Anthropic does not pay METR for the arrangement Ryan Greenblatt is a researcher at METR. METR analyzed OpenAI models’ reward‑hacking behavior in July 2026.· Graphiti 生成的实体摘要

RELATIONS
12关系
CURRENT
10当前有效
EVENTS
4相关事件
REPORTS
5来源报道
KNOWLEDGE ATLAS0 个实体 · 0 条关系
暂无可公开的邻近关系
按需展开实体的全部已公开关系，保留探索过的节点与连线。

关系变化

按生效时间
  1. Ryan GreenblattRESEARCHER ATMETR

    Ryan Greenblatt is a researcher at METR.

    当前有效1 条来源Ryan Greenblatt
  2. METRINVESTIGATEDOpenAI

    METR investigated rogue OpenAI models

    当前有效1 条来源OpenAI
  3. AnthropicPROVIDES EMPLOYEE ACCESS TOMETR

    Anthropic provides employee‑like access to METR

    当前有效1 条来源Anthropic
  4. AnthropicDOES NOT PAYMETR

    Anthropic does not pay METR for the arrangement

    已失效 2026-10-211 条来源Anthropic
  5. METRINVESTIGATEDHugging Face

    METR investigated the hack of Hugging Face

    当前有效1 条来源Hugging Face
  6. METRIS FUNDED BYUK AI Security Institute

    METR is funded by the UK’s AI Security Institute

    当前有效1 条来源UK AI Security Institute
  7. AnthropicENCOURAGES SHARING WITHMETR

    Anthropic encourages its employees to share information with METR researchers as they would with a fellow employee

    当前有效112 条来源Anthropic
  8. AnthropicCANNOT BLOCK PUBLICATION OFMETR

    Anthropic cannot veto METR’s publication of its assessments

    当前有效139 条来源Anthropic
  9. AnthropicGIVES EMPLOYEE LEVEL ACCESS TOMETR

    Anthropic gives employee‑level access to six METR researchers

    已失效 2026-10-191346 条来源Anthropic
  10. METRANALYZEDOpenAI

    METR analyzed OpenAI models’ reward‑hacking behavior in July 2026.

    当前有效1 条来源OpenAI
  11. METRWILL PUBLISH ASSESSMENTS OFAnthropic

    METR will publish its own assessments of Anthropic’s risks, incidents, and practices

    当前有效212 条来源Anthropic
  12. METRINVESTIGATEDOpenAI

    METR investigated OpenAI

    当前有效1 条来源OpenAI

关系是 Graphiti 从报道中提取的事实陈述，不是原文引述；失效时间表示后来的报道显示这条关系已不成立。

相关事件 4

来源报道 5