Skip to content
Moojaz — موجز
Search
Menu
HomeRegionsMaghrebAfricaMiddle EastEuropeAsiaAmericasWorldSectionsPoliticsEconomySecurityHealthScienceTechnologyAIEnvironmentCultureSportsSearch
United States of America
File

Anthropic research acceleration and external assessments

Published Updated

This dossier follows external assessments of AI use in Anthropic research, distinguishing measurements on defined tasks, evaluator judgments and demonstrated autonomous capabilities.

Latest event : 1 key moments
Editorial principlesReport an error
anthropic logo.
Anthropic logo, identification illustration. Public domain; PNG rendering. — Anthropic, Wikimedia Commons — domaine public (PD-textlogo).

Where things stand

On 5 November 2026, METR estimates that progress on well-specified tasks has roughly doubled in pace. Attribution to AI and generalization to other tasks remain uncertain.

Browse the full timeline

Reported information

METR estimates Anthropic's progress has doubled in pace on well-specified tasks

Independent evaluator METR published an assessment on 5 November 2026 estimating that Anthropic's capabilities on well-specified tasks are probably improving about twice as fast as between March 2024 and October 2025. The estimate carries substantial uncertainty.

anthropic logo.
Anthropic logo, identification illustration. Public domain; PNG rendering. — Anthropic, Wikimedia Commons — domaine public (PD-textlogo).

METR reviewed internal data and interviewed employees from 12 August to 29 October. It judges that AI use in research probably explains most of the acceleration, without establishing definite causation. Researchers consider that actual acceleration could be closer to 1.3 or 3 times the earlier pace.

The report mainly describes assistance to researchers, rather than their complete replacement by autonomous systems. Anthropic redacted some evidence important to the findings, limiting public verification of the conclusions.

Sources 3