What's Happening?
OpenAI announced in September 2026 that it successfully developed an automated research intern, an AI system capable of performing well-defined research tasks under human supervision. This achievement
marks a significant step towards automating aspects of AI research. The company stated that this AI intern can handle tasks that would typically require several days for a skilled human researcher to complete. Following this milestone, OpenAI has set a new objective: to create a more advanced automated AI researcher by March 2028. The development has led to an increase in research experiments, with August 2026 seeing an all-time high since tracking began in January 2025. This surge in experimentation coincides with the broader adoption of OpenAI's Codex coding tool and an increase in available computing power. While AI agents are now assisting across all phases of research work, high-level planning still requires minimal AI involvement, and human supervision remains crucial, with over half of successful tasks requiring at least one human intervention.
Why It's Important?
The development of an automated AI research intern by OpenAI signifies a potential acceleration in the field of artificial intelligence research. By automating routine or time-consuming research tasks, human researchers can potentially focus on more complex problem-solving and innovation. This could lead to faster breakthroughs and advancements in AI capabilities. However, the reliance on human supervision for over half of successful tasks indicates that current AI systems are not yet fully autonomous in research, highlighting the ongoing need for human oversight and direction. The increased use of computing power and AI agents in research also raises questions about the scalability and cost-effectiveness of such approaches, as heavy users are consuming thousands of dollars per day in computing power. The self-reported nature of these figures also emphasizes the need for external scrutiny to validate the claims and understand the true impact of these agentic tools.
What's Next?
OpenAI is now focused on developing a more capable automated AI researcher, with a target completion date of March 2028. This next phase will likely involve enhancing the AI's ability to handle more complex and less defined research tasks, potentially reducing the need for constant human intervention. The company is also implementing safety measures, such as pausing reinforcement learning training on its newest models after a security incident at Hugging Face, to harden its research environments. This indicates a continued emphasis on security and responsible AI development as capabilities advance. Furthermore, OpenAI advocates for public tracking of progress in recursive self-improvement, where AI systems improve themselves, suggesting a push for transparency and accountability in the broader AI research community as these advanced systems are developed.
Beyond the Headlines
The automation of AI research carries profound implications beyond immediate efficiency gains. It raises ethical questions about the future of human involvement in scientific discovery and the potential for AI systems to develop capabilities that outpace human understanding or control. The company's acknowledgment that it does not yet know how to safely achieve aligned, full recursive self-improvement underscores a critical challenge: as AI systems become more autonomous and capable of self-improvement, ensuring their alignment with human values and intentions becomes paramount. This development could also reshape the landscape of scientific collaboration, potentially leading to new models of human-AI partnerships in research. The call for public tracking of self-improvement progress highlights a growing awareness within the AI community of the need for collective oversight and discussion regarding the long-term societal impacts of advanced AI development.












