OpenAI has disbanded its standalone robotics team again, moving researchers into broader divisions under Mark Chen and Greg Brockman. The company says general-purpose AI models can increasingly support robotics tasks, although experts stress the continuing importance of real-world training data.

OpenAI Disbands Robotics Team for Second Time, Shifts Focus to General-Purpose AI

The420 Staff
6 Min Read


Follow us on Google News


Follow The420.in on Google News

OpenAI has disbanded its standalone robotics division for the second time in six years, moving researchers and hardware specialists into its broader research and applied organisations as the company changes its approach to developing general-purpose robots.

The restructuring, confirmed by OpenAI, comes as the ChatGPT maker increasingly believes that its advanced artificial intelligence models can perform tasks previously requiring specialised robotics systems.

Under the new arrangement, robotics researchers have been reassigned to the main research organisation led by Chief Research Officer Mark Chen. Applied robotics and hardware teams will operate within the company’s Applied division, headed by President Greg Brockman.

OpenAI maintains that the organisational changes do not represent an exit from robotics. Instead, the company says integrating robotics expertise with its broader AI research will help accelerate progress towards machines capable of performing a wide range of physical tasks.

FCRF’s Chief AI Officer Certification Returns on November 14 to Prepare Future AI Leaders

Why Has OpenAI Disbanded Its Robotics Division Again?

OpenAI previously operated a dedicated robotics research team before winding it down in 2020.

Co-founder Wojciech Zaremba later explained that limited availability of robot training data and the potential of large language models had influenced the company’s decision to pursue other approaches to artificial intelligence.

OpenAI returned to robotics in 2024, establishing another standalone team and partnering with humanoid robotics developer Figure to build AI models for robots.

However, the partnership ended in February 2025 after Figure decided to develop its own robotics AI technology.

The latest restructuring reflects another change in OpenAI’s research strategy.

According to the company, its researchers initially expected advances in robotics to emerge partly from technologies developed for Sora, its video-generation model, which focused on understanding physical environments and interactions.

OpenAI now believes its newer frontier models, including GPT-6 Astra, have developed capabilities that can be applied directly to robotics.

This has prompted the company to integrate robotics researchers into teams working on general-purpose AI rather than maintain a separate division.

Can General-Purpose AI Replace Specialised Robotics Models?

The decision has prompted discussion among robotics researchers about whether increasingly capable foundation models could reduce the need for specialised AI systems.

Guanya Shi, an assistant professor at Carnegie Mellon University’s Robotics Institute, said the restructuring suggests robotics may increasingly become an application of general intelligence rather than a separate research challenge.

He noted that general-purpose models are becoming capable of handling variations in physical tasks, such as changes in object positions or lighting conditions, that previously required specialised training.

However, Shi also emphasised the continued importance of smaller models that can operate directly on robots without requiring constant internet connectivity.

Russ Tedrake, a professor at the Massachusetts Institute of Technology and chief executive of Walden Robotics, highlighted another limitation of relying solely on general-purpose AI.

He said robotics companies possess an advantage through data collected from machines already operating in factories and other workplaces.

Such real-world experience allows developers to train systems for practical tasks and environments, providing information that general AI research alone may not capture.

Pete Florence, co-founder of robotics AI startup Generalist, also questioned whether any single approach to developing machine intelligence would dominate the robotics industry.

The differing assessments suggest that general-purpose models may improve robots’ capabilities without eliminating the need for specialised hardware, local processing and real-world training.

Robotics Ambitions Continue Amid Organisational Changes

OpenAI has maintained that the restructuring is intended to strengthen rather than discontinue its robotics programme.

The company is developing its own humanoid robot, with Chief Executive Sam Altman targeting a first public demonstration in 2027.

The reorganisation also comes as OpenAI faces pressure to concentrate resources on its principal products and research priorities.

The company has scaled back certain initiatives while seeking to manage substantial operating costs and growing demand for computing infrastructure.

However, OpenAI has not described the latest restructuring as a decision to abandon robot development.

Its researchers will continue working on robotics-related capabilities within the wider research and applied divisions.

The move reflects OpenAI’s belief that increasingly capable general-purpose AI models can serve as a foundation for multiple applications, including software development, digital assistants and physical robots.

Whether those models can reliably handle the complexities of real-world robotics without extensive specialised training remains a central question for the industry.

About the author — Suvedita Nath is a science student with a growing interest in cybercrime and digital safety. She writes on online activity, cyber threats, and technology-driven risks. Her work focuses on clarity, accuracy, and public awareness.