OpenAI says it has reached an “automated research intern” milestone
6 September 2026
OpenAI says an internal agent can complete well-defined research tasks that would take a skilled researcher several days. It reports 3.1 agent-workdays for each human research workday by mid-August, while stressing that the measurements are preliminary and researchers still set priorities and decide whether to scale, pause or deploy.
Why it matters: This is one of the clearest public disclosures of agents moving from coding assistance into the AI-research loop.
Impact: Enterprises should treat internal productivity metrics as hypotheses, require independent evaluations, and place research agents behind scoped compute, tool and approval controls.
Company claim only; the publication is verified, but internal metrics are not independently audited. Medium
