OpenAI will have a fully automated AI researcher by 2028: Sam Altman
OpenAI’s CEO has set 2028 as the goal for a true automated AI researcher, with an ‘AI research intern’ targeted for September 2026; the company has also said it has become a public benefit corporation overseen by the OpenAI Foundation.
OpenAI’s chief executive Sam Altman has set March 2028 as the goal for delivering a ‘true automated AI researcher’, stating the target during a livestreamed Q&A on 28 October 2025.
He also outlined an intermediate milestone: an ‘automated AI research intern’ by September 2026.
Altman described the 2028 objective as a practical, testable goal that is more useful than debating definitions of artificial general intelligence (AGI). In his remarks, he said OpenAI intends to demonstrate concrete research capabilities rather than pursue a label.
What is a ‘legitimate AI researcher’?
During the session, Altman described a system that can autonomously generate and test original hypotheses, elevating AI from a research assistant to a researcher in its own right.
OpenAI’s on-stage discussion has placed emphasis on the model’s ability to handle larger, multi‑step projects with minimal human hand‑holding.
Altman’s comments have come as OpenAI has said it has completed a restructuring: the OpenAI for‑profit entity has become a public benefit corporation overseen by the OpenAI Foundation, which continues to control the company.
If OpenAI delivers on the 2026 and 2028 milestones, the resulting systems would be judged by their performance on real research workflows. That outlook aligns with a broader push in the research community to evaluate AI as scientific assistants through domain‑grounded tasks and experiments, rather than generic benchmarks.