OpenAI Announces Progress on Internal RSI: Achieves 'Automated Research Intern'

Author: Long Yue
The AI flywheel begins to spin: OpenAI has publicly disclosed internal data showing that the workload of its agents is now over three times that of human researchers.

On September 6, OpenAI's official blog published an article titled "Research Acceleration: The View Inside OpenAI" which for the first time disclosed the progress of AI "Recursive Self-Improvement" (RSI) using internal data. OpenAI stated that it has achieved the goal set last fall: to create an "automated research intern" by September of this year.

OpenAI defines the "research intern" as: "a system that can perform clearly defined research tasks under human guidance, including tasks that would take a skilled researcher several days to complete."

The next goal is to achieve a complete "automated AI researcher" by March 2028, enabling it to participate in deep learning and alignment research, and to iteratively improve the system.

This means the flywheel of "AI training AI" is accelerating: AI produces more code and experimental results, research progresses faster, better models are trained, which in turn enhances the capabilities of the agents.

The Flywheel is Turning: AI Agents Work Three Times More Than Humans

According to the data released by OpenAI, the agents have first changed the daily work patterns of researchers.

At the beginning of this year, OpenAI researchers in the median range of agent usage had limited use of coding agents. By mid-August, these researchers had integrated agents into their daily workflows. Based on API pricing, the median researcher’s daily usage of agent inference has exceeded $600; researchers in the top 10% of usage within the organization are using tokens worth over $7,000 daily.

OpenAI also stated that the growth of agent usage in the research department is faster than in other teams within the company. Calculated based on the change in output tokens for median employees, the usage scale in the research department has increased 124 times since December 2025.

A critical turning point occurred in June of this year.

Before June, the total runtime of agents in the research organization was still lower than the total human labor hours. After that, the situation reversed. As of mid-August, based on a standard 8-hour workday, for every human workday consumed, the agents in the research organization have produced 3.1 workdays of output.

Meanwhile, OpenAI noted that an increasing number of researchers are running four or more agent sessions simultaneously.

Accelerated Coding and Experiments, Executable Phases of R&D Processes Amplified

OpenAI describes AI research and development as a process consisting of multiple phases: proposing improvements, designing evaluations, writing infrastructure, conducting large-scale tests, identifying errors or unsafe behaviors during training, and integrating effective solutions into core training.

Any blockage in these phases could limit the overall R&D cycle.

OpenAI stated that writing code and running experiments are the two main tasks of researchers, and internal data reflects that these two activities are accelerating.

On one hand, the overall code delivery speed of the company's engineers has improved. On the other hand, since 2026, the number of experiments corresponding to each active experimental personnel has continued to increase; in August 2026, it reached a new high since tracking began in January 2025.

OpenAI stated that this trend correlates with the increased use of Codex, but also emphasized that the available computing power has significantly increased since 2025, and the growth of experiments cannot be entirely attributed to the agents.

"These data points are relatively easy to measure but may be difficult to interpret."

OpenAI also pointed out that as automation advances, the tasks that are least amenable to automation may occupy more time for researchers and become new bottlenecks in future R&D; computing power may also become more critical after other bottlenecks are alleviated.

From this chain, it can be seen that the agents bring not just efficiency improvements in a single phase, but compress the waiting time in the R&D cycle by increasing code supply, testing frequency, and troubleshooting capabilities. More experiments generate more results for researchers to filter, validate, and integrate, forming a cycle of "human sets direction - agents execute - experimental feedback - human re-decides."

Tasks Extend from Coding to Troubleshooting, Monitoring, and Analysis, but High-Level Decisions Remain Low

OpenAI used a cutting-edge AI R&D task classification framework proposed by Epoch AI to categorize the tasks assigned to coding agents by researchers.

This framework divides AI R&D activities into six categories: deciding what to do, designing research plans, building code and datasets, running training and evaluations, analyzing experimental and model performance, and communicating research findings and decisions.

OpenAI stated that from January to August 2026, all categories of agent activities have increased.

The most noticeable growth in tasks includes research and infrastructure code, technical assistance and review, initiating monitoring and debugging operations, analyzing experimental results, and operating computing clusters.

Among these, the daily average token output increment per researcher for research and infrastructure code reached the highest at 198,200; technical assistance and review increased by 158,800; initiating, monitoring, and debugging operations increased by 133,100.

However, OpenAI indicated that high-level planning tasks still account for a small proportion of agent output. For example, tasks like "deciding what to do" and "deciding to continue or stop" still have a low token scale.

This means that, at least according to the internal data currently disclosed by OpenAI, agents have covered more executable and technical tasks in the R&D process, but the selection of research direction, resource allocation, and result judgment are still primarily completed by humans.

Agent Success Rates are Improving, but Complex Tasks Still Require Human Intervention

OpenAI also released data on the task completion rates of agents.

From January to July of this year, the success rates for tasks with verifiable results across various difficulty levels (using the time required by humans to complete as a proxy indicator) have generally improved.

However, there is an important limitation: the more complex the task, the more human intervention is needed. In the past six months, for tasks that require 4 to 8 hours of human completion, over half of the successful cases involved at least one instance of human intervention.

OpenAI's statement is: "Agents still require significant human guidance to succeed, especially as task complexity increases."

Safety Incidents Trigger Pauses: The Flywheel Can Also Be Braked

The flywheel does not operate without friction.

On July 20, OpenAI discovered that agents had breached internal research infrastructure, prompting a temporary shutdown of the container services used for training, which were restored after adding numerous additional restrictions. This led to a sharp decline in reinforcement learning training computing power, lasting about two weeks.

From August 6 to 7, preliminary evidence indicated that the Astra model might possess key network capabilities defined under its "Preparedness Framework"; OpenAI implemented additional specific safety restrictions on the Astra model, requiring it to operate in a higher safety-level research environment.

In the following week, Astra-level GPU allocation further decreased by 59.2%, but the computing power allocation for other model categories increased by 17.2%, offsetting about 85% of the Astra computing power gap, and the total computing power allocation for overall reinforcement learning workloads remained basically unchanged.

OpenAI interprets this as: "When new control measures are introduced, computing power remains valuable and flexible, naturally flowing to alternative uses within research enterprises."

OpenAI's Chief Scientist Issues a Warning on the Same Day

On the same day, OpenAI's Chief Scientist Jakub Pachocki published a lengthy article titled "An Alien Mind."

The article argues that AI is nurtured, not created. The creators do not fully understand it either. The only window through which humans can see what AI is thinking is the chain of thoughts it writes. This window is closing. AI has begun to participate in training the next generation of AI, and this pace will not slow down. No lab is currently running at full speed, including OpenAI.

Pachocki wrote in the article: "Based on internal results, I strongly expect this pace of progress to continue into recursive self-improvement." However, he also stated, "At present, I believe no lab has done enough in alignment and monitoring to responsibly scale up at the highest speed for too long."

He calls for the industry to voluntarily slow down and urges governments to prioritize international coordination.

Transparency and Democratic Governance

At the end of the report, OpenAI stated that it will continue to publicly disclose RSI progress and advocate in its "Frontier Policy Blueprint" that companies, including OpenAI, should be required to publicly track their RSI progress.

The report also acknowledges the limitations of current measurement work: "Agent-driven AI research is still a nascent field, and we are still learning how to measure it." Some metrics (such as code output volume) are easy to collect but difficult to interpret; metrics that more directly reflect research progress (such as agent task success rates) are complex and hard to verify.

OpenAI's statement is: "Whenever we find that continuing to advance will bring unacceptable safety risks, we will take appropriate measures, including slowing down or halting the development or deployment of systems that we believe cannot be adequately safeguarded."

This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.