- Published on
Very Good Model, But Essentially Decelerationism, Hindering Further AI Capital Expenditure
- Authors
- Name
- DeepHorus
OpenAI's Strategic Future Lead Dean W. Ball recently publicly criticized Moonlight's open-source Kimi K3 model, calling it 'essentially decelerationism' that would hinder further capital expenditure in the AI field. This statement quickly drew industry attention, reflecting deep divisions between Chinese and American AI open-source approaches and regulatory strategies.
Kimi K3 is a flagship model released by Moonlight on July 16, with 2.8 trillion parameters. It topped the Frontend Code Arena global AI large model leaderboard with a score of 1679, surpassing Claude Fable 5. Moonlight has announced that it will release the model weights soon, continuing its open-source tradition. Dean W. Ball admitted that Kimi K3 'is a very good model' that cannot simply be attributed to techniques like distillation, but then turned to criticize the open-source behavior.
Ball pointed out that open-weight models are essentially 'decelerationism,' hindering further capital expenditure, and promoted the so-called 'open-source danger theory'—claiming that in the future, a 'non-biological, intangible, dangerous, and infinitely self-replicating agent' could escape from the laboratory. He also publicly suggested that the U.S. government 'does not need evidence' and could use institutions like the Federal Reserve to issue soft regulations stating that 'Chinese models may have backdoors,' creating regulatory panic to suppress competitors.
Public information shows that Dean W. Ball is an American AI policy expert who just joined OpenAI this month as Head of Strategic Futures. He previously served as a senior advisor at the White House Office of Science and Technology Policy. His remarks are seen as OpenAI's new stance on balancing AI safety and competition, especially targeting Chinese open-source models.
Analysts believe that Ball's criticism reflects concerns among some U.S. policymakers and businesses about open-source models accelerating AI proliferation, but selectively ignores the fact that open source promotes innovation and transparency. Moonlight has not yet responded to Ball's comments.
The open-sourcing of Kimi K3 may further promote the development of domestic large model ecosystems, but it may also heighten vigilance among U.S. regulators. Industry observers point out that AI competition is extending from the technical level to the governance level, and future games will be more complex.
