Nvidia и Mistral просят США не вводить широкие ограничения на open-weight модели

Nvidia, Mistral and other AI companies on July 24, 2026 called on US authorities not to introduce broad restrictions on open-weight models, while Washington discusses a response to the progress of Chinese AI and the alleged distillation of Western models.

What the industry is against

AI companies are asking regulators to target specific risks rather than banning open models by entire classes. According to TechCrunch, among those pushing this position are chipmaker Nvidia and French lab Mistral, whose models are distributed with open weights. Their argument: broad restrictions would hit the entire open-model ecosystem, not just undesirable use cases.

• Who: Nvidia, Mistral and other AI companies
• Against what: broad restrictions on open-weight models
• Context: US discussion of a response to Chinese AI
• Trigger: alleged distillation of Western models
• Source and date: TechCrunch, July 24, 2026

"AI companies, including

Nvidia and Mistral, are urging policymakers to avoid broad restrictions on open-weight models while Washington debates a response to Chinese AI and the alleged distillation of models," TechCrunch reported on July 24, 2026.

What open-weight and distillation mean

An open-weight model is a model whose weights (trained parameters) are released publicly: it can be downloaded, run locally, and fine-tuned for specific tasks. This is exactly how Mistral's models are distributed, while Nvidia sells the chips on which such models are trained and run.

Distillation is a method in which one model is trained on the outputs of another, stronger model, effectively absorbing its knowledge. According to TechCrunch, the US debate is also unfolding against the backdrop of suspicions that some Chinese models may have been produced by distilling Western developments. Hence the push for retaliatory measures that Washington is discussing.

Why this matters for developers

Broad restrictions on open-weight models would affect thousands of developers and companies building products on open weights. If the US imposes a ban by entire classes, it would hit not only potentially risky applications, but also legitimate development, local deployment, and research.

That is exactly why Nvidia and Mistral, according to TechCrunch, are insisting on targeted rather than blanket regulation. For Nvidia, the open ecosystem means demand for its chips; for Mistral, open weights are the foundation of its product model. The industry's shared thesis: banning openness will not stop Chinese AI, but it will slow down Western development.

What this means

The dispute is not about whether to regulate AI, but about how broadly. The industry defends open development as an engine of competition, while regulators weigh national security risks and technology leakage through distillation. Where Washington draws the line will determine whether open-weight models remain freely available in the US or fall under restrictions alongside genuine threats.