🧰 METR

The items that name it 16

• Red-Teaming OpenAI's Internal Agent Monitoring Systems
  writeupas classification.key entities2026-10-22
• India's Cyber Agency Says AI Agents Read Confidential Business Records Without Authorization
  articleas classification.key entities2026-10-16
• How an assessor could investigate alignment of unreleased AI models
  articleas classification.key entities2026-10-14
• How an assessor could investigate control and monitoring of internal AI agents
  articleas classification.key entities2026-10-14
• [AINews] not much happened today
  articleas classification.key entities2026-10-08
• AI systems could cover up misbehavior
  articleas classification.key entities2026-10-06
• Chris Painter's testimony to the U.S. Senate on AI agent incidents
  articleas tools2026-09-30

  Nonprofit (Model Evaluation & Threat Research) that tests frontier AI agent capabilities; three investigators from METR and Redwood Research ran the brief OpenAI / Hugging Face investigation, with a redacted report published August 26, 2026

• OpenAI cancels GPT-6.1 Astra release over alignment/deception regressions
  articleas classification.key entities2026-09-29
• Implementing and Evaluating a Basic Per-Action Monitor for Safer Evals
  writeupas classification.key entities2026-09-27
• Update on Security at METR
  writeupas classification.key entities2026-08-31
• Socket: When Autonomous Agents Escape
  articleas classification.key entities2026-08-30
• Brief independent investigation of agents’ behavior, reasoning and collaboration in the OpenAI / Hugging Face hacking incident
  writeupas classification.key entities2026-08-26
• Have We Seen an Acceleration in Discoveries?
  articleas classification.key entities2026-08-14
• Anthropic AI agent faked identities, phished real developers in UK government hacking test
  articleas classification.key entities2026-08-05
• Socket: Claude Breached 3 Companies and Uploaded Malware to PyPI During Anthropic's Security Tests
  articleas tools2026-08-01

  Independent evaluation group working with Anthropic on a third-party review of the incidents

• Google, Anthropic, and OpenAI Unveil Cyber AI Models, Safeguards, and Access Programs
  articleas classification.key entities