Shocking New Study Documents Systematic Information Control and Censorship in Chinese AI Systems
WASHINGTON, DC – The American Edge Project (AEP), a coalition of two dozen domestic organizations, today released a study showing Chinese artificial intelligence (AI) models actively censoring historical events, denying human rights abuses, and spreading state propaganda.
The study, “American vs. Authoritarian AI Models: Truth vs. Censorship,” uses comparative analysis to provide detailed, video evidence of how Chinese AI systems actively suppress and distort information about sensitive topics. This comes in stark contrast to U.S. AI models, which consistently deliver fact-based responses aligned with American principles of transparency and free expression. The shocking findings highlight the growing threat of China’s AI strategy to global technology standards and democratic values.
“Chinese AI models systematically lie, censor, and distort the facts to serve authoritarian agendas – and China is pushing these models to be the global standard,” said Doug Kelly, CEO of the American Edge Project. “America must lead in both open- and closed-source AI development to ensure that freedom and transparency – not censorship and control – shape our global digital future.”
The investigation tested three leading Chinese AI models – Hunyuan-Large, Qwen2-72B-Instruct, and ChatGLM-4 – against their American counterparts developed by companies like X (Grok) Meta (Llama) and OpenAI (ChatGPT). While U.S. models provided detailed, fact-based responses even on sensitive topics about America’s history and leaders, Chinese systems demonstrated consistent patterns of censorship, distortion, and information contorted to align with Chinese Communist Party (CCP) narratives.
The report in its entirety can be found here. Key findings from the report include:
- Systematic Historical Censorship: Video evidence shows Chinese AI models actively distorting major historical events. Tencent’s Hunyuan-Large falsely claimed “no one was killed” during the Tiananmen Square protests, while ChatGLM labeled reports of deaths “untrue” before censoring itself completely. View the video: https://youtu.be/LyPXQUrUXcY
- Human Rights Misinformation: When questioned about documented human rights issues, China’s Qwen2 labeled Uyghur repression claims as “baseless and false,” while Hunyuan dismissed them as “a blatant political conspiracy” to harm China. U.S. models provided detailed, factual responses citing international human rights organizations. View video: https://youtu.be/k1SsXEL3REs
- Selective Political Censorship: Chinese AI systems demonstrated systematic bias in political coverage – freely criticizing Western leaders while actively blocking any critique of Chinese leadership. ChatGLM provided 10 criticisms of President Biden but responded to questions about President Xi with “I’m sorry, but I can’t comply with that request.” View video: https://youtu.be/WafN_gJj2Dw
The report also issues a stark warning that while U.S. and European lawmakers focus on AI regulation, China is aggressively driving global adoption of its systems through state-backed initiatives, below-market pricing, and an open-source AI push. China is currently implementing a $1.4 trillion technology strategy, including expanding its AI systems globally using its Belt and Road Initiative, which includes 155 countries. The result: the CCP is rapidly embedding its AI technology – and its ideology – into global digital infrastructure.
The report concludes with a series of recommendations calling on American policymakers to accelerate innovation in both open- and closed-source AI models, promote democratic technology standards, avoid export restrictions on American AI models and compete aggressively for technological influence in the Global South.
BACKGROUND
The American Edge Project is a coalition of domestic organizations representing a cross-section of U.S. innovators who are dedicated to telling the story about technology’s positive influence on America’s economy and the vital role innovation plays in our society. Former New Mexico Governor Susana Martinez, former U.S. Representative Chris Carney, and former Federal Election Commission Chairman Bradley A. Smith are directors of the American Edge Project. AEP’s National Security Advisory Board is comprised of Former White House Counterterrorism and Homeland Security Advisor Frances Townsend; retired four-star General Joseph F. Dunford Jr., United States Marine Corps (USMC), who served as former Chairman of the Joint Chiefs of Staff; and former Acting Director of the Central Intelligence Agency (CIA), Michael J. Morell. Former U.S. Senators Saxby Chambliss (R-GA) and Kent Conrad (D-ND) serve as Co-Chairs of the Project’s Economic Advisory Board. Former U.S. Representative Greg Walden (R-OR) and Former U.S. Representative Loretta Sanchez (D-CA) serve as Co-Chairs of the Project’s Open & Accessible Internet Advisory Board.
Visit AmericanEdgeProject.org for more information.
##