A survey of 349 technical workers finds a median 1.4–2x self-reported change in value of work due to AI tools, expected to grow over time, though there are reasons to be skeptical of the magnitude.
External review from METR of the "Risks from automated R&D" section in Anthropic's February 2026 Risk Report
We distinguish three measures of AI uplift -- on old tasks, on new tasks, and in value -- and show that task substitution can cause these to diverge substantially.
Classifying human and agent contributions to the NanoGPT speedrun, and what publicly tracked challenges can tell us about AI R&D acceleration.
This is a linkpost for MirrorCode, a project that METR funded and co-developed with Epoch AI. See Epoch AI’s blog post for more detail: https://epoch.ai/blog/mirrorcode-preliminary-results/
We find that a small amount of fine-tuning on instruction following in the CoT generalizes to meaningful increases in CoT controllability on an out-of-distribution set of tasks. We fine-tune four reasoning models on small datasets of instruction-following...
A METR staff member spent three weeks red-teaming a subset of Anthropic's internal agent monitoring and security systems, discovering several novel vulnerabilities.
Alexander Barry examines how different modelling choices affect METR's time horizon estimates.
Thomas Kwa describes a tabletop exercise where METR researchers simulated having access to ~200-hour time horizon AIs.
External review from METR of Anthropic's Sabotage Risk Report for Claude Opus 4.6
We find that roughly half of test-passing SWE-bench Verified PRs written by recent AI agents would not be merged into main by repo maintainers. A naive interpretation of benchmark scores may lead one to overestimate how useful agents are without more...
Nikola Jurkovic describes observations from tasking Opus 4.6 with reimplementing Slay the Spire and Balatro in the CLI.