Artificial intelligence is rapidly becoming the main way Americans learn about the past, understand current events, and make decisions about the future. That shift has turned the competition between the United States and China into a battle over the information that powers AI systems.
China’s growing AI influence
Researchers have documented several ways political power can seep into AI. A 2026 study in Nature found that Chinese state‑coordinated media appears in major AI training datasets. When that material was added to an open‑weight model, the model’s answers grew more favorable toward Chinese political institutions and leaders.
Further tests by Stanford and Princeton scholars showed that Chinese‑origin models were more likely to refuse sensitive questions, give shorter answers, or provide inaccurate information about Chinese politics. In some cases, the models mischaracterized Chinese dissident Wei Jingsheng and Nobel laureate Liu Xiaobo.
Bias in AI outputs
Other research highlights bias that can emerge later in the AI pipeline. A study of the DeepSeek model across 646 politically sensitive topics found that sensitive information sometimes disappeared or was reformulated before reaching the user. A separate analysis of 36,000 political prompts showed that model origin and the language of the query affected responses on Chinese sovereignty and human‑rights issues.
Experiments presented at the 2025 Association for Computational Linguistics meeting demonstrated that users exposed to liberal‑biased or conservative‑biased models were more likely to adopt opinions aligned with the model’s bias, even when those opinions conflicted with their own political affiliation. A large Nature Communications study involving 4,829 participants found AI‑generated political messages could shift attitudes on issues such as semi‑automatic‑rifle bans, carbon taxes and paid parental leave.
American response under President Trump
President Trump has taken decisive action to protect the integrity of American AI. In 2025 he directed federal agencies to procure large‑language models that prioritize truth‑seeking, historical accuracy and ideological neutrality. He also established the American AI Exports Program to promote U.S. models, hardware, software and standards in allied markets.
These steps reflect a broader belief that free societies can expose, challenge and reverse political pressure on AI developers. Meta CEO Mark Zuckerberg recently recounted how senior officials in the previous administration pressured his company to remove certain COVID‑19 content, a pressure the President’s policy seeks to prevent by ensuring transparency and competition.
Domestic AI landscape
American AI developers are also addressing bias. xAI markets its Grok assistant as “truth‑seeking,” while Anthropic publishes a Constitution for its Claude model that explicitly shapes its values and limits. Such voluntary safeguards complement the administration’s push for truth‑focused procurement.
Experts argue that without a single authority dictating permissible answers, free societies retain the ability to test models, switch products and build alternatives. The Trump administration’s policies aim to preserve that flexibility while safeguarding the nation’s informational environment.
What lies ahead
The United States must continue to lead in advanced AI research, energy, chips, computing power, capital and talent. Improving visibility into training data and developing attribution tools will help identify state‑directed information. At the same time, the government must avoid granting any single agency the power to decide which historical account is true.
As AI moves deeper into schools, journalism, government, medicine and business, the stakes grow. The country that leads AI will shape how billions retrieve the past, understand the present and decide the future. A free society’s decisive advantage remains that no answer has to be final.
Original reporting: Fox News (HLL/CB) — read the source article.
