AivexaNewsSearch
AI news for builders and product teamsChecked every hour

China protests the US Deepseek ban while blocking ChatGPT, Claude, and Gemini at home

Collected Oct 21, 2026

The US Department of Homeland Security has ordered civilian federal agencies to strip products made by the Chinese AI company Deepseek from their IT systems. The order, reported by The Decoder, covers not only Deepseek's own software but also software from other vendors built on the company's open-source models, and it bars agencies from making future purchases. Washington's stated concern is that Beijing could reach sensitive data through the software.

China has responded sharply. The Foreign Ministry accuses the US of stretching the concept of national security beyond reasonable limits, and the Commerce Ministry has demanded that the ban be lifted while vowing to take what it calls resolute countermeasures. According to the report, Beijing also proposed its own export and licensing controls for AI technology last week, following a model similar to the American approach.

The protest is awkward given China's own treatment of foreign AI. ChatGPT, Claude and Gemini are not approved for use there. Since 2023 the Cyberspace Administration has been required to review AI services, which must comply with political guidelines, and that requirement applies to the whole market rather than just government agencies. State bodies and state-owned companies are expected to gradually swap foreign hardware and software for domestic products, and reports point to restrictions on iPhones for government officials. Measured against that, the US action is narrower in scope, targeting federal civilian agencies rather than an entire consumer market.

The American side has its own stated rationale beyond data access. China's 2017 National Intelligence Law obliges companies and citizens to support the country's intelligence services, which is the legal hook behind concerns about compelled data disclosure. The report also notes that Chinese models carry a political slant, avoiding certain subjects and repeating state messaging less critically than Western models. On the economic front, AI companies and the FBI are among those that have repeatedly accused Chinese firms of copying leading US models at industrial scale; Beijing denies this.

This suggests the exchange is best read as positioning rather than a dispute about consistent principles. Each government is invoking national security to shield its own market and technology base, and the US does something similar on the hardware side through strict chip export controls, which tie security policy to economic policy. The symmetry is loose but real: both sides are building legal instruments that can be widened later.

For developers, the practical questions are contractual and architectural. The US order reaches beyond Deepseek's own products to third-party software built on its open-source models, so a federal civilian agency, or a contractor supplying one, may need an inventory of which dependencies trace back to Deepseek weights or code, not just which vendors appear on a purchase order. That is a harder audit than checking a logo on a procurement form, because open-weight models can be redistributed and fine-tuned under many names. A likely trade-off is that teams will start treating model provenance the way they treat software supply-chain provenance, with bills of materials and documented origins. It is an inference, but a reasonable one, that procurement reviews will ask for that evidence.

For product teams outside government, the immediate effect is limited in the US, since the order applies to civilian federal agencies. The more consequential signal is the tit-for-tat dynamic. China's promise of countermeasures, combined with its reported move toward export and licensing controls on AI technology, hints at possible restrictions on outbound technology or on foreign vendors operating in China. Companies with AI operations or partners on both sides should treat regulatory exposure as a two-way risk rather than a one-directional compliance cost.

Why it matters: Developers and product teams building on open-weight Chinese models now face a provenance question, because the US ban extends to third-party software derived from Deepseek's models, and government-facing work may require tracing dependencies rather than vetting vendors by name. Companies with Chinese operations or partnerships should also watch for reciprocal restrictions, since Beijing has promised countermeasures and has floated its own export and licensing controls; that reading of where this leads is inference, not a stated outcome. The episode reinforces that model choice is increasingly a geopolitical decision, not only a technical one.

Read at The Decoder

Based on reporting from the original publisher. Visit the source for full context and later updates.

Publisher excerpt

The US has banned Deepseek from its government systems, and China is protesting sharply and demanding the ban be lifted. Yet Beijing itself blocks ChatGPT, Claude, and Gemini across its entire market. China's outrage isn't very credible, but it should be taken seriously as a political signal of possible retaliation. The article China protests the US Deepseek ban while blocking ChatGPT, Claude, and Gemini at home appeared first on The Decoder .