Alibaba's Qwen3.8 re-educated with a more Western worldview
With the help of a post-training process, Qwen3.8-27B has been reborn under the name Arete "to reinforce Western values on a Chinese-trained model." What qualifies as a "Western value" remains open to interpretation.
- The company also notes that its efforts to apply "unlearning" to Qwen3.6-35B-A3B saw a dramatic drop in output that reflects Communist Party of China's preferred censorship, propaganda messaging, or bias.
- "On our evaluation, Arete-27B raised its fact score from 2.32 to 2.98 out of 5 and cut answers containing propaganda from 22 percent to 1 percent, in line with the Western models we tested," the company said.
- Some Middle Kingdom models have been found to produce more vulnerable code under certain conditions, according to a study by Booz Allen Hamilton, which does a lot of business with the US government.
What's perhaps more surprising is that Anthropic claims operators affiliated with Alibaba's Qwen lab used outputs from its Claude models to train competing AI models via distillation.
An American AI lab has re-educated Qwen3.8-27B, a popular open-weight large language model from Alibaba, so the software forgets its Chinese ancestry.
- Nous says the risk posed by Chinese-made AI models go beyond promoting Beijing's worldview.
- Your correspondent tested this with a variant model, Qwen3.8-27B-MLX-6bit, locally.
- The concern comes amid broader scrutiny of Chinese influence in the tech industry, heightened by Nvidia CEO Jensen Huang's resignation following controversy involving a woman US intelligence has accused of spying for China.
- "We have experienced similar contamination, because the vast majority of high quality open models are Chinese, and only open models have terms of use that permit this kind of tinkering and experimentation." In the end, the biz tapped Inkling, an open weight model from US-based Thinking Machines Lab, to act as the teaching model.
- Nous Research says its model re-education efforts show that the propaganda layer is shallow and, with 50 million tokens worth of supervised learning, is relatively easy to remove.
How this summary was made: extractive summary — built from the sources' own reporting. It is drawn only from the reporting of the sources listed below — read the originals for full detail. How we summarize →