Report Reveals Leading AI Models Echo Chinese Propaganda, The Register Discovers

A recent report by the American Security Project highlights bias in five popular AI models, which promote viewpoints aligned with the Chinese Communist Party (CCP) while censoring material deemed unfavorable. The models examined include OpenAI’s ChatGPT, Microsoft’s Copilot, Google’s Gemini, DeepSeek’s DeepSeek-R1, and X’s Grok. Despite only one model being based in China, all showed tendencies to reflect CCP propaganda, especially when discussing sensitive topics like the Tiananmen Square massacre. Microsoft’s Copilot was noted for presenting CCP narratives as credible, while Grok was more critical. The report emphasizes that AI models lack an inherent understanding of truth, instead generating responses based on statistical patterns from their training data. The authors call for more rigorous data training standards to prevent the dissemination of disinformation, urging greater caution in using these technologies, given their potential to propagate harmful ideologies or biases.

Source link