Trump’s “anti-ai” order could reshape the way in which American technological companies form their models

When Deepseek, Alibaba and other Chinese companies have published their AI models, Western researchers quickly noticed that they have bypassing the Chinese Communist Party. US officials later confirmed That these tools are designed to reflect Beijing’s discussion points, which raises concerns about censorship and prejudices.

US IA and OPENAI leaders have stressed this as a justification to advance their technology quickly, without too much regulation or surveillance. As the OpenAi World Affairs Director wrote, Chris Lehane Linkedin Post Last month, there is a competition between “democratic AI led by the United States and the Chinese Autocratic AI led by the Communists”.

An executive decree Signed Wednesday by President Donald Trump who prohibits the “awakened AI” and AI models which are not “ideologically neutral” from government contracts could disrupt this balance.

Order calls for diversity, equity and inclusion (DEI) calling it an ideology “omnipresent and destructive” which can “distort the quality and precision of production”. More specifically, the order refers to information on the race or sex, the manipulation of racial or sexual representation, the critical theory of race, transgender, unconscious bias, intersectionality and systemic racism.

Experts warn that it could create a frightening effect on developers who may feel pressure to align the model outputs and data sets with white house rhetoric to guarantee federal dollars for their cash activities.

The order comes the same day that the White House published Trump AI action plan», Which takes the national priorities from the societal risk and focuses on the construction of AI infrastructure, the reduction of administrative formalities for technological companies, national security and competition with China.

The order directs the director of the Office of Management and Budget as well as the Federal Policy Administrator, the General Services Administrator and the Director of the Office of Science and Technology Policy, to issue advice to other agencies on how to comply.

“Once for all, we get rid of Woke,” said Trump on Wednesday during an AI event organized by the all-in podcast and Hill & Valley Forum. “I will sign an order prohibiting the federal government from obtaining AI technology which has been imbued with partisans or ideological agendas, such as the critical theory of race, which is ridiculous. And from now on, the American government will only deal with AI which pursues the truth, equity and strict impartiality. “

To determine what is impartial or objective is one of the many challenges to order.

Philip Seareant, lecturer in linguistics applied to open university, told Techcrunch that nothing can be objective.

“One of the fundamental principles of sociolinguistics is that language is never neutral,” said the sergeant. “So the idea you can get pure objectivity is a fantasy.”

In addition to that, the ideology of the Trump administration does not reflect the beliefs and values of all Americans. Trump has repeatedly sought to eliminate the financing of climatic initiatives, education, public broadcasting, research, social services subsidies, community and agricultural support programs and affirmative care, often supervising these initiatives as examples of “awakened” or politically biased government spending.

While Rumman Chowdhury, a data scientist, CEO of non -profit technology Humane Intelligence, and former envoy of American sciences for AI, said: “All that the Trump administration does not like) is immediately thrown into this derogatory pile from Woke.”

The definitions of “research of truth” and “ideological neutrality” in the order published on Wednesday are vague in some respects and specific in others. While “research of truth” is defined as LLMs which “prioritize historical precision, scientific research and objectivity”, “ideological neutrality” is defined as LLM which are “neutral and non -partisan tools which do not manipulate the answers in favor of ideological dogmas such as Dei”.

These definitions leave room for a large interpretation, as well as potential pressure. AI companies have pushed less constraints on how they work. And although a decree does not carry the strength of the legislation, the companies of the border AI could still find themselves subject to the changing priorities of the political agenda of the administration.

Last week, Openai, Anthropic, Google and Xai signed contracts With the Ministry of Defense to receive up to $ 200 million each to develop agental workflows that solve critical national security challenges.

We do not know which of these societies is best placed to gain the ban on awakened AI, or if they comply.

Techcrunch has contacted each of them and will update this article if we hear.

Despite the biases of bias, XAI can be the most aligned on order – at least at this early stage. Elon Musk positioned Grok, Xai’s chatbot, like the ultimate anti-alarm clock, “less biased”, the truth of truth. The guests of the Grok system ordered him to avoid postponing the authorities and the traditional media, to seek information against even Reference of musk opinions On controversial subjects. In recent months, Grok has even marked out anti -Semitic comments And rented Hitler on X, among other hateful articles, racists and misogynists.

Mark Lemley, professor of law at the University of Stanford, told Techcrunch that the decree was “clearly intended for the discrimination of sight, since (the government) has just signed a contract with Grok, alias” mechahitler “.

In addition to the financing of the DOD of XAI, the company announced that “Grok for the government“Had been added to the calendar of the general services administration, which means that XAI products are now available for purchase in all government offices and agencies.

“The right question is as follows: do they prohibit Grok, the AI with which they have just signed a large contract, because it has been deliberately designed to give politically loaded answers?” Lemley said in an e-mail interview. “Otherwise, it is clearly designed to discriminate a particular point of view.”

As shown by the prompts of the Grok system, the model outputs can reflect the two people on technology and the data on which AI is formed. In some cases, an overabundance of prudence between developers and AI formed on Internet content which promotes values such as inclusiveness has led to distorted model outputs. Google, for example, last year fell under fire After his Chatbot Gemini showed a black George Washington and racial Nazis – which Trump’s order calls an example of AI models infected by Dei.

Chowdhury says that his greatest fear with this decree is that AI companies will actively rethink data training to tow the party line. She underlined affirmations From Musk a few weeks before the launch of Grok 4, saying that Xai would use the new model and its advanced reasoning capacities to “rewrite the whole corpus of human knowledge, adding missing information and deleting errors. Then recycle yourself. “

This would ostensibly place musk in the position of judging what is true, which could have huge implications downstream on how the information is accessible.

Of course, companies have made judgments on the information that the information is seen and not seen since dawn of the Internet.

Conservatives like David Sacks – the entrepreneur and the investor that Trump appointed as IA tsar – was frank about his concerns about “Woke Ai” on the all -in podcast, which co-organized Trump’s announcements from Trump. Sacks accused creators of important AI products of infusing them left-wing values, of framing its arguments in defense of freedom of expression and a warning against a tendency to a centralized ideological control in digital platforms.

The problem, say the experts is that there is no truth. It is impossible to obtain impartial or neutral results, especially in the world of today where even the facts are politicized.

“If the results that an AI produced says that climate science is correct, is it left-wing?” According to Searegeant. “Some people say that you must give both sides of the argument to be objective, even if one side of the argument has no status.”