Anthropic reported that its Claude system led 26% of internal AI research and development work in August, up from 1% in March. The company said roughly 30,000 AI agents were active on its main research platform, while humans collaborated on more than 90% of measured work. No measured activity was fully autonomous, but about one in 47,000 agent decisions was blocked before execution. Anthropic also said AI-assisted research devoted 12% of computing power to safety work in a sampled week.
Source: https://wireless.reuters.com/business/anthropic-says-claude-now-leads-quarter-work-building-its-next-ai-models-2026-09-17
