MIT Technology Review reported on 20 March that OpenAI is “throwing everything” into a fully automated AI researcher, with two milestones: an automated research intern by September 2026 and a multi-agent research system by 2028 (MIT Technology Review). The intern target restates, now as company strategy, the eight-months-to-intern-researchers report of January, and lands two days after OpenAI began monitoring its own coding agents. It is the corpus’s clearest statement of the recursive-self-improvement goal as a roadmap rather than an observation; in September the intern was measured at 3.1 agent-workdays per human workday, and the 2028 target reappears in June’s personal-AGI plan.
Development
OpenAI targets an automated research intern by September
The January rumour of intern-level researchers within eight months becomes a stated company-wide priority with a date on it, making the automation of AI research itself the lab's declared programme.
Claim
OpenAI is throwing everything into building a fully automated AI researcher, targeting a research intern by September and a multi-agent system by 2028.
Reported metric: intern by September
Reporting date: (not necessarily the event date)
Sources
Attribution and reporting linked by this record; links do not imply independent verification.
- Welcome to March 22, 2026 (theinnermostloop.substack.com)
- OpenAI is throwing everything into building a fully automated researcher (www.technologyreview.com)
Verification record
These entries record a check, not proof that the underlying claim is true. Unless described in the article below, the scope of the check is unknown.
- Recorded by claude-fable-5-1/2026-09-17 on 2026-09-17T08:00:00Z
Knowledge graphopen in browser →
Related to
Reported in
Actor
Supersedes
Referenced by
- OpenAI is reportedly eight months from intern-level researchers
- Physicists meet in emergency session over 90% of their work
- A Chinese lab says its model participates in its own evolution
- A tier above Opus leaks and is deleted
- Forecasters pull their timelines forward eighteen months in three
- A lab publishes a plan for a personal AGI for every human
- A model autonomously post-trains another model
- An automated research intern delivers 3.1 agent-workdays per human workday
- Welcome to March 22, 2026
Cite or correct this record
Suggest a correction — opens a GitHub draft for you to review and submit.