Andrew here. Silicon Valley is locked in a heated debate over new open-weight A.I. models out of China, with critics claiming they were “distilled” — read: stolen — from U.S. counterparts. Against this backdrop, the Trump administration and Chinese officials are expected to discuss A.I. regulation ahead of President Xi’s meeting with Trump on Sept. 24.

我是安德鲁。硅谷正陷入一场关于来自中国的新型开放权重人工智能模型的激烈辩论，批评人士声称这些模型是从美国同类模型中蒸馏而来的——换句话说，就是偷的。在此背景下，预计在9月24日习近平主席与特朗普会晤之前，特朗普政府与中国官员将就人工智能监管问题进行讨论。

The big question: Will Washington move to block Chinese open-weight models from the U.S.? And will this become a new chess piece in trade negotiations?

最大的问题是：华盛顿是否会采取行动阻止中国开放权重模型进入美国？这是否会成为贸易谈判中的新棋子？

A.I. protectionism

人工智能保护主义

The U.S.-China race for artificial intelligence dominance has shifted drastically in recent weeks, as more Chinese models (that are largely open-weight) mostly matched the best Western offerings.

最近几周，美中争夺人工智能主导地位的竞赛发生了急剧变化，因为更多中国模型（主要是开放权重模型）在性能上已基本可与西方顶尖产品相媲美。

New reports suggest that Washington and Beijing are weighing steps to protect and advance their national industries — which many in the industry believe could pose more problems than they solve.

新的报道表明，华盛顿和北京正在权衡保护和推进本国产业的措施——业内许多人认为，这些措施带来的问题可能比解决的问题还要多。

Some in the Trump administration are pushing for limits on foreign models. The emergence of powerful new open-weight Chinese models from Z.ai and Moonshot AI, which are cheaper to use than closed Western models, have reinvigorated efforts to effectively ban such tools, Axios reports, citing unnamed sources:

特朗普政府中的一些人正在推动对外国模型实施限制。据Axios援引未具名消息人士的报道，智谱AI和月之暗面推出的强大新型中国开放权重模型使用成本低于封闭的西方模型，这重新激发了旨在实际上封禁此类工具的努力：

The Commerce Department last year considered adding multiple Chinese A.I. labs to its “Entity List,” which would effectively cut off U.S. access without a license, a source close to the administration told Axios.

一位接近政府的消息人士告诉Axios，商务部去年曾考虑将多个中国人工智能实验室列入其“实体清单”，这实际上将切断未经许可的美国企业对这些实验室的访问权限。

The National Security Agency and White House Office of the National Cyber director also considered putting out an advisory on Chinese A.I. lab threats last year, practically discouraging U.S. companies from using their tech, the source said.

该消息人士称，美国国家安全局和白宫国家网络总监办公室去年曾考虑发布一份关于中国人工智能实验室威胁的警告，实际上是劝阻美国公司不要使用相关技术。

These voices have gained power in the administration, especially as critics of the approach — including David Sacks, the former White House A.I. czar; Sriram Krishnan, a top White House A.I. adviser; and Chris Fall, the director of the Commerce Department’s Center for A.I. Standards and Innovation — have departed.

随着这种做法的批评者纷纷离职——包括前白宫人工智能沙皇戴维·萨克斯、白宫人工智能高级顾问斯里拉姆·克里希南以及商务部人工智能标准与创新中心主任克里斯·福尔，这些声音在政府中获得了更大的影响力。

• Leaders of A.I. giants, including Dario Amodei of Anthropic and Sam Altman of OpenAI, have pushed for mandatory government testing and even licenses for frontier models, in the name of safety. Some have also argued against open-source models.

· 包括Anthropic的达里奥·阿莫迪和OpenAI的萨姆·奥特曼在内的人工智能巨头领导者以安全为名，一直推动政府对前沿模型实施强制性测试，甚至要求为前沿模型颁发许可证。一些人也曾公开反对开源模型。

Prominent industry figures are still pushing back, arguing that — contra the views of Amodei and Altman — open-source A.I. is good because it forces leading labs to compete on capability and price.

业内知名人士仍在提出异议。与阿莫迪和奥特曼的观点相反，他们认为，开源人工智能是有益的，因为它迫使领先的实验室在能力和价格方面展开竞争。

• “Lobbyists are urging Washington to treat open-model A.I. as a security threat. In fact, it is something more familiar: proper competition that should be welcomed,” Bill Gurley, the veteran venture capital investor, wrote in an opinion piece in The Washington Post

· “游说者正敦促华盛顿将开放模型的人工智能视为安全威胁。事实上，这是一种更常见的情况：是应该受到欢迎的正当竞争，”资深风险投资家比尔·格利在《华盛顿邮报》的专栏文章中写道。

• “The leading closed labs, already a duopoly in terms of A.I. model revenue, want the government to eliminate their open source competition,” Sacks wrote on X.

· “在人工智能模型收入方面已形成双头垄断的顶尖封闭实验室希望政府为他们消除开源的竞争对手，”萨克斯在X平台上写道。

China is reportedly weighing limits for its own models. The country’s commerce ministry has been considering ways to limit foreign companies from getting access to key A.I. and chip-making data or from acquiring important tech start-ups, The Financial Times reports, citing unnamed sources.

据报道，中国正在考虑对其本土模型实施限制。据《金融时报》援引未具名消息人士的报道称，中国商务部一直在考虑采取措施，限制外国公司获取关键的人工智能和芯片制造数据，或阻止它们收购重要的科技初创企业。

What to watch: The U.S. and China are set to hold talks in September over how to regulate their increasingly powerful models, according to Reuters.

关注焦点：据路透社报道，美中定于9月就如何监管两国日益强大的人工模型举行会谈。