Symbolic imageChinese censorship reported in US AI answers as DeepSeek targets Claude Code
The Wall Street Journal reported on Thursday that Chinese censorship is leaking into answers from American AI models. A day earlier, DeepSeek publicised its push against Anthropic's Claude Code.
The Journal published the finding under the headline 'Chinese Censorship Is Leaking Into Answers From American AI'; only that headline is public, so the models examined, the evidence and the scale cannot be checked. The report names no company in the portion available. For Western developers the claim raises a supply-chain question, since material of Chinese origin is widely used in training data.
DeepSeek aims at Anthropic's coding tool
Bloomberg reported on Wednesday, 12 August, that DeepSeek is publicising its efforts to challenge Claude Code, Anthropic's coding assistant. Global Times, in a note alongside its own reporting, said the company had released a next-generation model, the DeepSeek-V4 Preview.
Supernode for two-trillion-parameter models
Global Times reported the same day that China's first supernode built for two-trillion-parameter models had entered service in Ulanqab. The state outlet illustrated the piece with a photo of a 'Lingjun Zhenwu M890' supernode instance credited to Alibaba Cloud.
Tencent backs a new lab
Bloomberg also reported on Wednesday that a former Alibaba AI architect secured backing from Tencent and HSG for a new AI lab.
Other opinions
The two US outlets frame Chinese AI as risk and competition, while Global Times reports only capacity and product launches and does not mention the censorship finding.