Claude AI Drives 26% of Research and Development Work at Anthropic
Anthropic says its Claude assistant now leads 26% of the company’s internal AI research and development work, up from 1% in March. The company reported that AI systems collaborated with humans on more than 90% of development tasks in August, while thousands of agents operated under automated safety checks and oversight.
Artificial intelligence firm Anthropic has revealed that its Claude assistant now leads 26% of the company's internal research and development work. The disclosure forms part of a new transparency initiative designed to provide external observers with detailed metrics on how quickly advanced systems are contributing to the creation of future software generations.
As per a report by Reuters, the San Francisco startup noted that autonomous systems collaborated with human researchers on more than 90% of internal development tasks as of August. While Claude is not yet operating fully autonomously across complete workflows, its contribution has accelerated significantly from just 1% recorded in March on tracking metrics provided by Epoch AI. Anthropic Combines Claude Chat and Cowork Functions into Unified Interface with New Document Tools.
AI Agent Scaling and Safety Oversight
During peak operational periods in August, approximately 30,000 independent artificial intelligence agents simultaneously executed engineering and research tasks on the company platform. To maintain stringent operational boundaries, every action executed by the agents undergoes automated safety screening before implementation.
Company data indicates that out of over a billion individual decisions processed during the sample month, roughly one in 47,000 actions was actively blocked. Additionally, approximately 6 per cent of total computing power dedicated to research was allocated to safety evaluations, with that proportion doubling to 12 per cent for tasks conducted autonomously by AI systems.
Industry Transparency and Autonomous Development
The detailed metrics arrive amid broader discussions across the technology sector regarding the speed of artificial intelligence automation and the potential risks of systems growing increasingly self-sufficient. Industry researchers continue to monitor how leading developers balance rapid capability scaling with rigorous oversight mechanisms to prevent unexpected behavioural divergences. Anthropic IPO: AI Giant in Talks With Nvidia for USD 10 Billion Anchor Investment Ahead of Historic Listing
Anthropic intends to publish these performance and oversight statistics on a regular schedule to foster open dialogue. Independent analysts note that such disclosures offer vital visibility into the practical reality of recursive software self-improvement within premier artificial intelligence laboratories.
(The above story first appeared on LatestLY on Sep 18, 2026 08:53 AM IST. For more news and updates on politics, world, sports, entertainment and lifestyle, log on to our website latestly.com).