OpenAI's AI research interns are here, fully automated researchers by 2028

OpenAI says its AI research interns are now a reality, with agents already helping researchers write code and run experiments. The company is now aiming for a fully automated AI researcher by March 2028.OpenAI says its AI research intern goal is achieved.Last week's buzz in the AI world remains OpenAI’s GPT-6 Astra, which the company positioned as a major step towards AGI. Now, OpenAI revealed it has reached another milestone: AI research interns are here, and fully automated AI researchers could arrive by 2028.In a new blog post, OpenAI says it has achieved its goal of building an automated research intern that can work under human supervision. The company says these systems can handle well-defined research tasks that could take a skilled researcher a few days to complete.OpenAI's AI interns are already doing more workAI is already shaping the world, from academia to markets, in multiple ways. And now it seems to be the turn of researchers; it could drastically shape the future research space. OpenAI revealed its AI agents can now work as an extra hand for its research teams.The company says researchers are now using coding agents throughout the day, often running several sessions at the same time. These agents are writing more code, running more experiments and taking on increasingly complex tasks. By mid-August, OpenAI's research organisation was using the equivalent of 3.1 agent-workdays for every one human workday. The company also found that more researchers are running four or more agents at the same time.And the agents are not just being asked to write code anymore. OpenAI says researchers are delegating higher-level and longer-running tasks to them, including research planning, technical specifications, analysing experiments, monitoring training runs and troubleshooting research infrastructure. The results are not perfect, though. OpenAI says agents still need significant human guidance as tasks become more complicated. In the last six months, more than half of successful tasks estimated to take humans four to eight hours involved at least one human intervention.Fully automated AI researchers could arrive by 2028OpenAI says it is making “strong progress toward creating an automated AI researcher by March of 2028.” The idea is to build a system that can work on deep learning and AI alignment under human supervision, helping improve AI research itself.If that happens, AI would no longer be limited to helping humans write code or analyse experiments. It could increasingly take part in the research loop itself, coming up with ideas, building them, running experiments and analysing the results.OpenAI believes automated research could also help with AI safety and alignment, including developing better defences against increasingly capable AI systems.But the company is not pretending the road ahead is risk-free. It says more capable systems can become harder to monitor, and that safety and alignment work must scale alongside capabilities. OpenAI also says it is prepared to slow or stop development if a system cannot be sufficiently safeguarded.For now, humans are still deciding what research matters and which ideas are worth pursuing. But if OpenAI hits its 2028 target, the definition of an AI researcher could look very different by the end of the decade.- EndsPublished On: Sep 7, 2026 11:55 IST