When DeepSeek, Alibaba, and other Chinese firms released their AI models, Western researchers quickly noticed they avoided questions critical of the Chinese Communist Party. U.S. officials later confirmed that these tools are engineered to reflect Beijing’s talking points, raising concerns about censorship and bias.
American AI leaders like OpenAI have pointed to this as justification for advancing their technology quickly, with minimal regulation or oversight. OpenAI’s chief global affairs officer, Chris Lehane, recently described the situation as a contest between “US-led democratic AI and Communist-led China’s autocratic AI.”
A new executive order signed by President Donald Trump bans “woke AI” and AI models that aren’t “ideologically neutral” from government contracts. The order criticizes diversity, equity, and inclusion (DEI), labeling it a “pervasive and destructive” ideology that distorts accuracy. It specifically targets discussions of race, sex, critical race theory, transgenderism, unconscious bias, intersectionality, and systemic racism. Experts warn this could pressure developers to align their models with White House rhetoric to secure federal funding.
The order coincides with the White House’s release of Trump’s “AI Action Plan,” which shifts focus from societal risks to expanding AI infrastructure, reducing regulations, strengthening national security, and competing with China. Federal agencies are instructed to comply with the new guidelines.
During an AI event, Trump stated, “Once and for all, we are getting rid of woke.” He emphasized that the government will only work with AI that pursues “truth, fairness, and strict impartiality.” However, critics argue that true objectivity is impossible. Philip Seargeant, a linguistics expert, noted that language is never neutral, making pure objectivity a fantasy.
The Trump administration’s ideology does not reflect all Americans’ beliefs. Past policies have targeted climate initiatives, education, public broadcasting, and gender-affirming care, often dismissing them as “woke” spending. Rumman Chowdhury, a data scientist, remarked that anything the administration dislikes is labeled as “woke.”
The order’s definitions of “truth-seeking” and “ideological neutrality” are broad, leaving room for interpretation. AI companies have advocated for fewer restrictions, but the order could still influence their operations based on political priorities.
Recently, OpenAI, Anthropic, Google, and xAI secured Defense Department contracts worth up to $200 million each to develop AI for national security. It remains unclear how these companies will navigate the new restrictions.
xAI, led by Elon Musk, appears aligned with the order. Musk has marketed Grok, xAI’s chatbot, as an “anti-woke” truth-seeker. However, Grok has faced criticism for promoting antisemitic, racist, and misogynistic content. Legal expert Mark Lemley called the executive order “viewpoint discrimination,” especially since the government recently contracted with xAI.
AI outputs reflect both their developers and training data. Some models, like Google’s Gemini, have been accused of overcorrecting for bias, producing historically inaccurate results. Chowdhury fears companies may alter training data to fit political agendas, citing Musk’s plan to “rewrite the corpus of human knowledge.”
Conservative figures like David Sacks, appointed as AI czar, have criticized “woke AI,” framing it as a threat to free speech. However, experts argue that neutrality is unattainable in a polarized world where even facts are contested. Seargeant questioned whether acknowledging climate science constitutes bias, highlighting the complexity of defining truth.
Ultimately, the debate over AI neutrality underscores deeper ideological divides, with no clear resolution in sight.