What does Ben Thompson propose in 'Who's Afraid of Chinese Models?'

Thompson's Stratechery essay proposes two US legal changes. First, that Congress or the courts explicitly recognize collecting data for model training as fair use. Second, that terms of service which forbid distilling outputs of a paid model be made unenforceable for American developers. He argues both changes would let US open-weights efforts compete with Chinese labs like Alibaba and Moonshot on equal ground.

Which Chinese models does the essay focus on?

The essay focuses on the recent wave of open-weights releases from Chinese labs, including Alibaba's Qwen 3.8 Max Preview and Moonshot's Kimi K3, both landed in July 2026. Thompson also notes that Alibaba's move to release Qwen 3.8 Max as open weights lines up with Xi Jinping's recent remarks encouraging open source, openness, collaboration, and sharing as national strategy.

Is 'Who's Afraid of Chinese Models?' free to read?

The Stratechery post 'Who's Afraid of Chinese Models?' has a free public preview on stratechery.com, published 2026-07-20. The full essay and its podcast version are gated behind the Stratechery Plus subscription. Simon Willison quoted from and linked to the piece on his blog the same day, which gives a summary of the core argument without a subscription.

How does the piece frame US vs China AI competition?

Ben Thompson argues that the US strategy of tightening ToS restrictions and export controls is backfiring: it slows American open-source development without stopping Chinese labs, which now ship the leading open-weight models. In his framing, the fight for AI leadership is now less about who builds the biggest closed model and more about which country's open-weights ecosystem attracts developers.

Stratechery · 2026-07-20 · major

Ben Thompson: Who's Afraid of Chinese Models? — legalize training, allow distillation

Ben Thompson argues the US should recognize model training as fair use and prohibit terms of service that ban distillation, so American open models can compete with Chinese ones like Kimi K3 and Qwen 3.8 Max.

Ben Thompson's Monday essay proposing a two-line US legal fix so American open models can match Chinese ones.

Quick facts

Author	Ben Thompson
Publication	Stratechery
Published	2026-07-20
Access	Free preview; full article for Stratechery Plus
Format	Monday essay + podcast

What is it?

'Who's Afraid of Chinese Models?' is Ben Thompson's Monday essay for Stratechery, published July 20 in response to the July wave of Chinese open-weights releases — Moonshot's Kimi K3 and Alibaba's Qwen 3.8 Max Preview. Thompson uses those releases as the setup for a policy proposal aimed at Washington.

How does it work?

The essay lays out two concrete legal changes. First, treat data collection for AI training as fair use so US labs can train on the same corpora Chinese labs already use. Second, void terms of service that prohibit distillation, freeing American developers to build smaller open models by learning from paid US frontier APIs. Thompson connects the timing to Xi Jinping's recent public backing of open source as Chinese national strategy.

Why does it matter?

US export controls and increasingly restrictive frontier-model ToS aim to keep leading capabilities out of Chinese hands, but the past month showed the opposite: China now ships the strongest open-weight models. Thompson's argument reframes the policy conversation — rather than tightening walls further, the US should tear down its own domestic restrictions so American open models can catch up. It is likely to shape ongoing debate in DC over the AI Diffusion Framework and open-source safe harbors.

Who is it for?

AI policy watchers, US and Chinese lab strategists, open-source AI advocates

Frequently asked questions

What does Ben Thompson propose in 'Who's Afraid of Chinese Models?': Thompson's Stratechery essay proposes two US legal changes. First, that Congress or the courts explicitly recognize collecting data for model training as fair use. Second, that terms of service which forbid distilling outputs of a paid model be made unenforceable for American developers. He argues both changes would let US open-weights efforts compete with Chinese labs like Alibaba and Moonshot on equal ground.
Which Chinese models does the essay focus on?: The essay focuses on the recent wave of open-weights releases from Chinese labs, including Alibaba's Qwen 3.8 Max Preview and Moonshot's Kimi K3, both landed in July 2026. Thompson also notes that Alibaba's move to release Qwen 3.8 Max as open weights lines up with Xi Jinping's recent remarks encouraging open source, openness, collaboration, and sharing as national strategy.
Is 'Who's Afraid of Chinese Models?' free to read?: The Stratechery post 'Who's Afraid of Chinese Models?' has a free public preview on stratechery.com, published 2026-07-20. The full essay and its podcast version are gated behind the Stratechery Plus subscription. Simon Willison quoted from and linked to the piece on his blog the same day, which gives a summary of the core argument without a subscription.
How does the piece frame US vs China AI competition?: Ben Thompson argues that the US strategy of tightening ToS restrictions and export controls is backfiring: it slows American open-source development without stopping Chinese labs, which now ship the leading open-weight models. In his framing, the fight for AI leadership is now less about who builds the biggest closed model and more about which country's open-weights ecosystem attracts developers.

Try it

Read the free preview at stratechery.com