David Sacks, the White House's AI adviser, went public in July 2026 with a sharp rebuke of America's top AI labs, posting on X that OpenAI and Anthropic are pushing Chinese AI restrictions not out of patriotism but to protect what he called a "duopoly" in AI model revenue. That framing landed hard, because it came from inside the administration both companies have been lobbying.
The backstory goes back at least to 2025. OpenAI and Anthropic have been quietly pressing Washington to add Chinese AI entities to the Entity List, a designation that would bar American firms and individuals from doing business with them. Both labs already block users in mainland China and Hong Kong from their platforms, a self-imposed rule their competitors largely haven't followed. Anthropic went further, accusing Chinese firms including DeepSeek and Moonshot, the company behind the Kimi assistant, of systematically extracting knowledge from its models through deceptive methods, essentially reverse-engineering expensive research for a fraction of the cost.
The split inside Silicon Valley
By late June 2026, six Chinese AI models had shown up on major AI leaderboards, competing directly with Western systems. That fact alone made the restriction debate urgent. For startups and mid-size companies quietly running cost-effective Chinese open-source models in their products, a blanket Entity List designation would cut off a supply chain they've already built around.
The broader tech industry has pushed back hard. The argument is simple enough: blocking Chinese models reduces competition, inflates costs for American businesses that depend on affordable AI tools, and hands the two biggest labs a structural advantage they didn't earn through the market. Critics see the national security framing as a convenient wrapper around a commercial interest.
The Trump administration has been listening to both sides simultaneously, which has left policy in an unresolved state. OpenAI and Anthropic want formal restrictions. Sacks and much of the Valley want the opposite. The gap between those positions is wide, and neither side has moved.
What makes this harder to call is that the national security argument isn't entirely hollow. Distillation, if Anthropic's allegations hold up, represents a real shortcut for Chinese labs that could compress years of R&D into months. But the same open competition dynamic that worries the big labs is exactly what's been accelerating AI adoption across industries worldwide, often through cheaper, more accessible tools.
For now, the Entity List discussions are ongoing and no formal action has been announced. The two most valuable AI companies in the country want Washington to act. The White House's own AI adviser is telling them to back off.
