Experts say exploiting Anthropic’s Fable isn’t how Kimi K3 got so good
Quick Answer
Experts dispute claims that Moonshot's Kimi K3, a leading open-weight LLM, was developed through distillation of Anthropic’s Fable, citing insufficient time and resources.
Quick Take
U.S. officials allege that Moonshot used banned chips and proprietary technology, raising concerns over tech theft and export regulations.
Key Points
- Moonshot's Kimi K3 is the largest available open-weight .
- U.S. officials allege Moonshot copied Anthropic’s Fable LLM.
- Experts doubt distillation is responsible for Kimi K3's capabilities.
- Advanced Nvidia chips used by Moonshot are banned for export to China.
- Distillation practices are common across the AI industry, not just in China.
DeepSignal Analysis
What happened
Experts are questioning claims that Moonshot's Kimi K3, a prominent open-weight LLM, was developed through distillation of Anthropic's Fable. U.S. officials allege that Moonshot utilized banned chips and proprietary technology, raising concerns about potential technology theft and export violations.
Key evidence
- Michael Kratsios, a White House science advisor, stated that Moonshot built Kimi K3 by copying Anthropic's Fable while using chips not cleared for export to China.
- Experts like Braden Hancock argue that the rapid development of Kimi K3 cannot be solely attributed to distillation from Fable, citing insufficient time since Fable's public release on July 1st.
- Anthropic accused Moonshot of systematically distilling its models earlier this year, claiming to have identified millions of exchanges between its models and users at Moonshot through IP addresses.
Why it matters
The allegations against Moonshot highlight ongoing tensions between the U.S. and China in the AI sector, particularly regarding technology transfer and intellectual property. If proven, these claims could lead to stricter regulations on AI development and export controls, impacting both companies and research initiatives. The situation underscores the complexities of AI model development and the challenges of distinguishing between legitimate research and potential theft.
Source Excerpt
"I don't think you get a model this strong and this quickly on the heels of Fable doing strictly distillation," one expert told TechCrunch.
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from TechCrunch
See more →
Why the first GPU financiers are turning to inference chips in a $400 million deal
General Compute secured a $400 million loan from Upper90, using inference-specific chips as collateral, signaling a shift towards cost-effective AI infrastructure. Their SN50 chips promise 16x faster inference than traditional GPU clouds, highlighting a growing market for open-source AI models and alternatives to Nvidia.
