OpenAI alleges effort to copy its AI model's reasoning

OpenAI has announced it detected what it describes as an attempt to extract proprietary reasoning techniques from its AI models. The company linked a portion of this activity to Moonshot AI, a Chinese AI startup, though full details of the alleged scheme were not disclosed.

The disclosure comes amid intensifying competition between US and Chinese AI developers, with concerns growing over techniques like model distillation, where a newer AI system may be trained to mimic a more advanced one's outputs or reasoning patterns. OpenAI has previously raised similar concerns about other companies.

Neither OpenAI nor Moonshot AI has provided extensive public comment beyond the initial allegation, and the claims remain unverified by outside parties.