JoyIn CEO Accuses OpenAI of Distilling Robotics Model, Threatens Lawsuit Over 90% Success Claim
Updated
Updated · CNBC · Sep 11
JoyIn CEO Accuses OpenAI of Distilling Robotics Model, Threatens Lawsuit Over 90% Success Claim
3 articles · Updated · CNBC · Sep 11
Summary
JoyIn CEO Guo Renjie said Thursday the Ant-backed startup has begun preparing a lawsuit after accusing OpenAI of directly "distilling" its robotics model and copying space-themed design elements.
Guo said JoyIn presented its "extraterrestrial visitor" framework in Silicon Valley weeks before OpenAI Chief Scientist's Sept. 6 "An Alien Mind," citing overlaps in recursive self-improvement and AI-led compute optimization.
The accusation came as JoyIn publicly unveiled its Aether model, which it says uses a perceptive rather than text-based control approach to give humanoid robots a 90% first-attempt task success rate.
Developer Zhu Mingxuan, formerly of Figure, told CNBC earlier this week that Aether cut training time by two-thirds and that JoyIn plans to open-source touch-related components but not energy-use modules.
The clash lands amid a wider U.S.-China AI race and a heated debate over model distillation, after a U.S. cybersecurity agency this week said six Chinese companies had distilled models from Anthropic, Google and OpenAI.