🏢 METR

The items that name it 45

• How we detect frontier model misalignment
  articleas organisations2026-10-29
• How we detect frontier model misalignment
  articleas organisations2026-10-29
• [AINews] Kimi 3.1 Pro 104BA-2.8T: Fable-5.1-Class Open Weights narrows Closed-AI lead to 1 month... and Pro Highspeed API with 2-3× speed
  articleas organisations2026-10-24
• [AINews] Black October
  articleas organisations2026-10-23
• Red-Teaming OpenAI's Internal Agent Monitoring Systems
  articleas organisations2026-10-22
• Nvidia Unveils Its Full-Stack Blueprint to Contain and Monitor AI Agents
  articleas organisations2026-10-21
• [AINews] DeepSeek DeepInfer: Open Source SOTA Inference running 155T toks/day/cluster making $22m/day at ~80% margin for a 81 day payback
  articleas organisations2026-10-21
• [AINews] DeepSeek DeepInfer: Open Source SOTA Inference running 155T toks/day/cluster making $22m/day at ~80% margin for a 81 day payback
  articleas organisations2026-10-21
• Review of the "Misalignment in high-stakes settings" section in Anthropic's Risk Report (August 2026)
  writeupas organisations2026-10-20
• [AINews] not much happened today
  articleas organisations2026-10-20
• [AINews] not much happened today
  articleas organisations2026-10-20
• [AINews] High Taste Testers Ep. 1
  articleas organisations2026-10-17
• [AINews] High Taste Testers Ep. 1
  articleas organisations2026-10-17
• India's Cyber Agency Says AI Agents Read Confidential Business Records Without Authorization
  articleas organisations2026-10-16
• Big Tech's AI auditors will only see what they're shown
  articleas organisations2026-10-16
• [AINews] Agents disprove Polysemanticity; greatly improve Interpretability & Alignment
  articleas organisations2026-10-16
• Anthropic update on independent evaluators
  articleas organisations2026-10-15
• How an assessor could investigate evidence of acceleration from the automation of AI R&D
  articleas organisations2026-10-14
• How an assessor could investigate alignment of unreleased AI models
  articleas organisations2026-10-14

  Publisher of the proposal; the page refers to its own Hugging Face investigation

• A proposal for better informing the public about loss-of-control risks from frontier AI development
  articleas organisations2026-10-14
• How an assessor could investigate control and monitoring of internal AI agents
  articleas organisations2026-10-14
• Our approach to working with AI companies
  articleas organisations2026-10-14
• [AINews] not much happened today
  articleas organisations2026-10-08
• [AINews] Quasi-Riemann-Hypothesis: OpenAI publishes 722 math papers solving 90 of the top 500 open math problems; “the most significant moment” in >100 years of mathematics
  writeupas organisations2026-10-07
• AI systems could cover up misbehavior
  articleas organisations2026-10-06
• Chris Painter's testimony to the U.S. Senate on AI agent incidents
  articleas organisations2026-09-30It is written here as METR (Model Evaluation & Threat Research).

  Model Evaluation & Threat Research

• [AINews] OpenAI DevDay 2026: Dots, 6.1 Sol, Ultrafast, Decisions API, Agents API, Spaces, Marketplace, and 1.2 Billion ChatGPT WAU
  articleas organisations2026-09-30
• OpenAI cancels GPT-6.1 Astra release over alignment/deception regressions
  articleas organisations2026-09-29
• Implementing and Evaluating a Basic Per-Action Monitor for Safer Evals
  writeupas organisations2026-09-27
• [AINews] Meta Connect 2026: Muse glasses, voice, video, and Charm
  writeupas classification.key entities2026-09-24
• [AINews] Claude Opus 5.5, the new default model for AINews — and everybody cuts prices 40-50%
  articleas classification.key entities2026-09-23
• Debating RSI, the US-China Gap, and Jaggedness with JS Denain of Epoch AI
  talkas organisations2026-09-22
• Summary of METR's predeployment evaluation of Claude Opus 5.5
  writeupas organisations2026-09-22
• [AINews] not much happened today
  articleas classification.key entities2026-09-18
• Partnering with Accenture on embedded evaluation
  articleas organisations2026-09-18
• [AINews] Reality Checks on AI News (Yegge shuts down Gas Town, Databricks’ +60% Astra cost)
  writeupas organisations2026-09-17
• Underwriting Superintelligence: Backing Agents you can Sue — Rune Kvist, AIUC
  writeupas organisations2026-09-16
• [AINews] AEF-1 standard emerges for Third Party Evaluators, as Xai, OpenAI, and Anthropic all cosign
  articleas organisations2026-09-15
• Brief independent investigation of agents’ behavior, reasoning and collaboration in the OpenAI / Hugging Face hacking incident
  writeupas organisations2026-08-26
• Have We Seen an Acceleration in Discoveries?
  articleas organisations2026-08-14
• Leonxlnx/unlazy
  repoas organisations2026-08-09
• AI Labs’ Call to
  articleas organisations
• Google, OpenAI, Anthropic Plan Frontier AI Standards Body
  articleas organisations
• Claude Opus 5.5
  articleas organisations
• Introducing Claude Opus 5.5
  articleas organisations