200 startups urge Trump: don't ban China AI
The newly formed Little Tech Association, backed by Y Combinator and Proton, asked Trump officials not to impose a blanket ban on Chinese open-weight AI models.
Nearly 200 venture-backed startups, organized under the newly formed Little Tech Association and supported by Proton and Y Combinator, sent a coordinated letter this week urging President Trump, Commerce Secretary Howard Lutnick, and OSTP director Michael Kratsios not to impose a blanket ban on Chinese open-weight AI models such as those from Moonshot AI and Alibaba.
American leadership requires two things: world-leading American open-weight models and continued access for U.S. builders to open models already available worldwide.
The group’s core argument is practical rather than ideological: startups that build products on top of free, downloadable open-weight models would face higher costs if forced onto paid API access from the small number of dominant US labs instead, which the letter argues would shrink competition rather than protect it. It also makes a security point, that once model weights are downloadable and already circulating worldwide, banning American builders from using them doesn’t stop foreign actors from having them, it just puts US companies at a disadvantage. The group is asking for targeted safeguards instead of a blanket restriction.
The letter lands the same week the administration escalated its rhetoric on the other side of this fight, with Treasury threatening sanctions over allegations that Moonshot distilled Anthropic’s Fable to build Kimi K3. Whatever the merits of that specific claim, this letter is the clearest sign yet that a broad ban would face real pushback from the startup ecosystem that’s actually using these models, not just from the Chinese labs that make them. If price or capability has you routing to a Chinese open-weight model, the policy ground here is actively contested, not settled in either direction, so it is worth having a fallback path ready regardless of how the distillation dispute resolves.
Update, July 24: The pushback against a broad ban widened beyond startups. A follow-up open letter to policymakers is now signed by established companies including Hugging Face, Meta, Microsoft, Mistral, NVIDIA, and Replit, making a similar case:
Policymakers should be careful not to conflate legitimate model-development techniques with misappropriation.
The letter argues distillation is a widely used technique for improving models generally, not evidence of theft on its own. Replit CEO Amjad Masad put the stakes more bluntly:
Banning Chinese open models is as good as banning open models in general.
Notably absent from the letter are OpenAI, Anthropic, Google DeepMind, and SpaceX. Separately, security researchers cited by TechCrunch say the technical evidence doesn’t support the distillation theory driving Treasury’s sanctions push, meaning the underlying IP-theft claim behind this whole fight is now contested on the facts, not just the policy.