Claims That Kimi K3 Distilled an Unfinished Anthropic Model Remain Unsupported

A viral claim says Anthropic CEO Dario Amodei told Congress that Kimi K3 may have distilled an Anthropic model that had not yet been completed. The claim appears to stem from satire, while available reports attribute the broader allegation to White House technology official Michael Kratsios instead.

Published 8/3/2026, 12:13:24 AMLast updated 8/3/2026, 12:13:24 AMEditorial publication

No reliable primary document confirms that Amodei submitted such a report or made the quoted statement in a congressional hearing. Publicly available accounts also provide no disclosed technical evidence showing that Kimi K3 reproduced capabilities from an unfinished Anthropic model. The timeline alone makes the specific claim difficult to substantiate.

Some reports describe allegations that Moonshot AI collected large volumes of outputs from Anthropic systems and may have used restricted computing resources. Those reports acknowledge that supporting evidence has not been made public. Moonshot has reportedly rejected the allegations and attributed Kimi K3’s performance to its own architectural work.

The story should therefore be treated as an unverified model-provenance and regulatory dispute, not as an established case of model theft. Stronger conclusions would require official investigation records, forensic model analysis, or direct statements from the parties involved.

Source evidence 6

Source review does not guarantee permanent accuracy. Consider the original source, observation time, and later corrections.

Anthropic CEO质疑Kimi K3模型来源，业内关注中美AI竞争从技术走向规则博弈 - 腾讯云开发者社区-腾讯云developer.cloud.tencent.com · supporting

腾讯云开发者社区文档建议反馈控制台首页学习活动专区圈层工具 MCP广场文章/答案/技术大牛发布 # Anthropic CEO质疑Kimi K3模型来源，业内关注中美AI竞争从技术走向规则博弈文章来源：企鹅号 - 二三里资讯举报近日，Anthropic首席执行官达里奥·阿莫代伊（Dario Amodei）在美国国会听证会上表示，中国AI公司Moonshot AI（月之暗面）推出的Kimi K3模型可能通过模型蒸馏技术学习了Anthropic尚未公开发布的内部模型能力。不过，截至目前，Anthropic尚未公开披露支持这一说法的技术证据，月之暗面方面也未对此作出公开回应。与此同时，美国政府正讨论进一步限制部分中国人工智能模型在美国市场的使用，引发业内对于AI技术竞争和监管政策的广泛关注。近年来，美国人工智能产业形成了”更强模型推动更多算力需求、算力需求带动AI芯片增长”的发展模式，英伟达等算力企业也因此持续受益。而随着中国开源大模型快速发展，越来越多模型以较低训练和部署成本获得较强性能，行业竞争正由单纯的算力竞争逐步转向模型效率、开源生态和商业化能力的综合竞争。多位业内人士认为，围绕大模型的竞争已经不仅是技术路线之争，也涉及产业生态、知识产权、监管政策和国际规则等多个层面。未来，各国监管政策可能在一定程度上影响AI产品的国际推广，但长期来看，模型能力、开发者生态和应用落地仍将是决定市场竞争力的关键因素。业内分析认为，随着中国开源模型不断迭代，全球人工智能产业竞争正在进入新的阶段。技术创新、开放生态和产业化能力，或将成为全球AI企业竞争的新焦点。 [...] 发表于: 原文链接：腾讯「腾讯云开发者社区」是腾讯内容开放平台帐号（企鹅号）传播渠道之一，根据《腾讯内容开放平台服务协议》转载发布内容。如有侵权，请联系 [email protected] 删除。 0 分享分享快讯到朋友圈分享快讯到 QQ 分享快讯到微博复制快讯链接到剪贴板上一篇：传统养老短板凸显，数字化赋能康养服务全新升级下一篇：2026年工作记录仪定制生产服务商靠谱商家测评排名 ## 相关快讯 ### Kimi K3来了，中国大模型开始给硅谷出题 2026-07