OpenAI reveals how AI agents are accelerating its research
Research acceleration: The view inside OpenAI
OpenAI shares internal data showing that coding agents have become integral to its researchers' daily work, with agent usage exceeding human labor hours by August 2026. The company reports increased code output, experiment volume, and a shift toward delegating more complex tasks to agents. It also discusses its commitment to safety, including pausing training on models with potential cyber capabilities, and outlines progress toward an automated AI researcher by 2028.
In terms of a standard 8 hour workday, as of mid-August, in total, the research organization uses 3.1 agent-workdays of effort for every workday of human labor.