ThinkPatternGet the app
Story
TECHNOLOGY · NOV 19, 2026

AI Decision Models Exhibit Bias Toward Chinese Communist Party

A report warns that AI decision models used for hiring and moderation embed Chinese Communist Party biases, even when developed by Western companies.

A report from the Australian website The Strategist warns that a new class of AI decision models used for automated hiring and content moderation exhibits a marked bias toward the politics of the Chinese Communist Party. This bias persists in models developed by Western companies when they are built upon Chinese AI foundations, such as Alibaba's Qwen.

TypeSafe, a San Francisco-based company, developed the decision model Jev, which was adopted by 25 percent of Fortune 500 companies within two weeks of launch due to its efficiency. Evidence of bias in these models includes the tendency to remove jokes about President Xi Jinping and reject job candidates who support independence for Xinjiang.

The report indicates that companies are prioritizing price and availability over supply chain security to reduce costs. While firms like Microsoft and Cloudflare have begun building their own decision models using existing foundations, the report urges Australia and its democratic partners to require companies to disclose AI foundations and test for political bias.


Reported across 3 outlets
Actors
Alibaba GroupChinese Communist PartyMicrosoftCloudflare

Keep reading in the app

The full story and every source, free in the app.