White House said to weigh targeted bans on specific Chinese open AI models

The New York Times reports the US favors selective bans over a blanket restriction on Chinese open-weight models, citing security.

The White House favors targeted bans on specific Chinese open-weight AI models rather than a blanket ban, according to a New York Times report. The Trump administration is said to be preparing to tighten restrictions, with national security cited as the reason. Open-weight models are those whose parameters are published for anyone to download and run, and recent frontier examples have increasingly come from China.

The industry response is split and tangled. OpenAI and Google DeepMind signed an open letter opposing regulation of open-weight models, alongside many tech companies with stakes in the open ecosystem. Microsoft and Google sell access to open models, and DeepMind develops its own Gemma series. Yet the same report says Anthropic and OpenAI are privately lobbying to restrict Chinese models — a notable gap given both face price pressure from cheaper Chinese alternatives.

For access to cheap frontier intelligence, restrictions cut against abundance. Open weights let developers, researchers and small firms run capable models locally at low cost, without paying per-query fees to a handful of providers. Selective bans would gate that supply, tilting the market back toward paid Western APIs.

The security case is not empty: the report notes that models such as Kimi K3 trail leading Western systems on cybersecurity benchmarks. The tension is between locking down potential misuse and keeping low-cost capability broadly available. This is reported as a plan in development, sourced to the New York Times, so the scope and enforcement remain unsettled.

Source: The Decoder