OpenAI says its automated research intern milestone has arrived

OpenAI says it has reached the automated research intern milestone it set in 2025 for September 2026. The company describes a system able to contribute to deep-learning and alignment work under human supervision, presenting the achievement as progress toward an automated AI researcher.

The distinction matters because research work involves forming hypotheses, designing experiments, interpreting failures and deciding which results deserve further investment. Automating parts of that loop could increase the number of experiments a laboratory can run and shorten the time between model generations.

OpenAI argues that the public needs visibility into frontier systems’ capabilities, incidents and safeguards so their development can be governed democratically. Its article frames supervised automation as a way to advance both deep learning and alignment, but does not claim that oversight questions are solved. The milestone should therefore be read as the company’s measured internal assessment.