OpenAI reported that AI agents in its own research already perform a volume of work equivalent to 3,1 human workdays for every human workday. The company also stated it has achieved the goal of creating an "automated research intern." This is reported by The Decoder.
Simultaneously, OpenAI Chief Scientist Jakub Pachocki warned that no laboratory currently possesses sufficiently reliable control over the alignment of systems with human goals and their monitoring to scale pace without restrictions.
The practical implication of the report lies in the gap between the growing role of AI in research work and persisting control limitations. However, the source is presented only as metadata and a brief synopsis, so details of the claims and their verifiability remain unclear.
editorial commentary
Why it matters
The likely consequence is increased attention to measuring AI agent productivity alongside requirements for their monitoring. The next observable signal will be the publication of primary methodologies, task examples, or more detailed safety assessments. Significant uncertainty remains because currently only a synopsis from The Decoder is available without primary confirmation.