Suzhou-based humanoid robotics startup JoyIn has accused OpenAI of unauthorized model distillation, claiming the artificial intelligence giant mirrored its “extraterrestrial visitor” framework and website design. According to CNBC, JoyIn CEO Guo Renjie publicly alleged in a letter that OpenAI directly distilled his company’s technology without modifications following a Silicon Valley presentation held weeks prior to OpenAI Chief Scientist’s “An Alien Mind” publication on Sept. 6.
Intellectual Property Claims and Alleged Technical Overlaps
Guo raised questions regarding similarities in core technological approaches, pointing specifically to “recursive self-improvement” and the optimization of computing power using artificial intelligence. According to CNBC, the Suzhou-based executive also noted design parallels between the outer space-inspired layout of OpenAI’s GPT-6 Astra web page and JoyIn’s Aether model website, which Guo stated was released two months earlier. JoyIn has initiated the process of filing a lawsuit against OpenAI, though CNBC was unable to independently verify the startup’s claims. OpenAI did not immediately respond to a request for comment regarding the allegations.
Broader Industry Scrutiny Over Model Distillation
The friction between OpenAI and JoyIn unfolds amid heightened industry scrutiny regarding how AI developers train advanced systems. According to CNBC, U.S.-based Anthropic has repeatedly flagged unauthorized model distillation by Chinese firms seeking to enhance their own artificial intelligence capabilities. Furthermore, a U.S. cybersecurity agency stated on Tuesday that six Chinese companies, including DeepSeek and Alibaba, distilled models from major U.S. developers like Anthropic, Google, and OpenAI.
Contrasting industry perspectives highlight the legal and ethical gray areas surrounding the practice. Garry Tan, chief executive of startup accelerator Y Combinator, told CNBC this month that he would “do nothing” about distillation, joining critics who point out that U.S. AI companies frequently train their own foundational models on data covered under copyright law.
Inside JoyIn’s Aether Humanoid Control Model
JoyIn officially announced its Aether model on Thursday, positioning the framework as a perceptual rather than text-based approach to robotic control. According to CNBC, the system allows humanoid robots to complete tasks with a 90% success rate on the first attempt. Zhu Mingxuan, who led the development of Aether after leaving U.S. humanoid robotics firm Figure last year, told CNBC earlier this week that the model reduced training time by two-thirds. Zhu stated plans to open-source specific parts of the model related to touch, while keeping energy-use components proprietary.
Did you know? Getting humanoids and AI models to perform as intelligently as humans across a variety of tasks has remained a challenge, amid a broader tech race between U.S. and Chinese companies.
Frequently Asked Questions
What did JoyIn accuse OpenAI of doing?
According to CNBC, JoyIn CEO Guo Renjie accused OpenAI of unauthorized model distillation, alleging that OpenAI copied the technical framework and space-inspired web design of JoyIn’s Aether model.
How has OpenAI responded to the allegations?
OpenAI did not immediately respond to a request for comment from CNBC regarding JoyIn’s claims and threatened lawsuit.
What is the function of JoyIn’s Aether model?
According to CNBC, Aether utilizes a perceptive approach to robotic control designed to help humanoid robots complete physical tasks with a 90% success rate on the first attempt while reducing training time by two-thirds.
Stay Updated on the AI Tech Race
Explore our latest coverage on artificial intelligence developments, robotics breakthroughs, and global tech policy. Leave a comment below to share your perspective on these industry disputes.
Keep reading