Tech companies looking to sell their artificial intelligence technology to the federal government must now contend with a new regulatory hurdle: proving their chatbots aren\u2019t \u201cwoke.\u201d <\/p>\n
President Donald Trump\u2019s sweeping new plan to counter China in achieving \u201cglobal dominance\u201d in AI promises to cut regulations and cement American values into the AI tools increasingly used at work and home.<\/p>\n
But one of Trump\u2019s three AI executive orders<\/a> signed Wednesday \u2014 the one \u201cpreventing woke AI in the federal government\u201d \u2014 marks the first time the U.S. government has explicitly tried to shape the ideological behavior of AI.<\/p>\n Several leading providers of the AI language models targeted by the order \u2014 products like Google\u2019s Gemini and Microsoft\u2019s Copilot \u2014 have so far been silent on Trump\u2019s anti-woke directive, which still faces a study period before it gets into official procurement rules.<\/p>\n While the tech industry has largely welcomed Trump\u2019s broader AI plans, the anti-woke order forces the industry to leap into a culture war battle \u2014 or try their best to quietly avoid it.<\/p>\n \u201cIt will have massive influence in the industry right now,\u201d especially as tech companies are already capitulating to other Trump administration directives, said civil rights advocate Alejandra Montoya-Boyer, senior director of The Leadership Conference\u2019s Center for Civil Rights and Technology.<\/p>\n The move also pushes the tech industry to abandon years of work to combat the pervasive forms of racial and gender bias<\/a> that studies and real-world examples have shown to be baked into AI systems. <\/p>\n \u201cFirst off, there\u2019s no such thing as woke AI,\u201d Montoya-Boyer said. \u201cThere\u2019s AI technology that discriminates and then there\u2019s AI technology that actually works for all people.\u201d<\/p>\n Molding the behaviors<\/a> of AI large language models is challenging because of the way they\u2019re built and the inherent randomness of what they produce. They\u2019ve been trained<\/a> on most of what\u2019s on the internet, reflecting the biases of all the people who\u2019ve posted commentary, edited a Wikipedia entry or shared images online. <\/p>\n \u201cThis will be extremely difficult for tech companies to comply with,\u201d said former Biden administration official Jim Secreto, who was deputy chief of staff to U.S. Secretary of Commerce Gina Raimondo, an architect of many of President Joe Biden\u2019s AI industry initiatives. \u201cLarge language models reflect the data they\u2019re trained on, including all the contradictions and biases in human language.\u201d<\/p>\n Tech workers also have a say in how they\u2019re designed, from the global workforce of annotators<\/a> who check their responses to the Silicon Valley engineers who craft the instructions for how they interact with people. <\/p>\n Trump\u2019s order targets those \u201ctop-down\u201d efforts at tech companies to incorporate what it calls the \u201cdestructive\u201d ideology of diversity, equity and inclusion into AI models, including \u201cconcepts like critical race theory, transgenderism, unconscious bias, intersectionality, and systemic racism.\u201d <\/p>\n The directive has invited comparison to China\u2019s heavier-handed efforts to ensure that generative AI tools reflect the core values<\/a> of the ruling Communist Party. Secreto said the order resembles China\u2019s playbook in \u201cusing the power of the state to stamp out what it sees as disfavored viewpoints.\u201d<\/p>\n The method is different, with China relying on direct regulation by auditing AI models, approving them before they are deployed and requiring them to filter out banned content such as the bloody Tiananmen Square crackdown<\/a> on pro-democracy protests in 1989.<\/p>\n Trump\u2019s order doesn\u2019t call for any such filters, relying on tech companies to instead show that their technology is ideologically neutral by disclosing some of the internal policies that guide the chatbots. <\/p>\n \u201cThe Trump administration is taking a softer but still coercive route by using federal contracts as leverage,\u201d Secreto said. \u201cThat creates strong pressure for companies to self-censor in order to stay in the government\u2019s good graces and keep the money flowing.\u201d<\/p>\n The order\u2019s call for \u201ctruth-seeking\u201d AI echoes the language of the president\u2019s one-time ally and adviser Elon Musk, who has made it the mission of the Grok chatbot made by his company xAI. <\/p>\n But whether Grok or its rivals will be favored under the new policy remains to be seen. <\/p>\n Despite a \u201crhetorically pointed\u201d introduction laying out the Trump administration\u2019s problems with DEI, the actual language of the order\u2019s directives shouldn\u2019t be hard for tech companies to comply with, said Neil Chilson, a Republican former chief technologist for the Federal Trade Commission.<\/p>\n \u201cIt doesn\u2019t even prohibit an ideological agenda,\u201d just that any intentional methods to guide the model be disclosed, said Chilson, head of AI policy at the nonprofit Abundance Institute. \u201cWhich is pretty light touch, frankly.\u201d<\/p>\n Chilson disputes comparisons to China\u2019s cruder modes of AI censorship.<\/p>\n \u201cThere is nothing in this order that says that companies have to produce or cannot produce certain types of output,\u201d he said. \u201cIt says developers shall not intentionally encode partisan or ideological judgments.\u201d<\/p>\n With their AI tools already widely used in the federal government, tech companies have reacted cautiously. OpenAI on Thursday said it is awaiting more detailed guidance but believes its work to make ChatGPT objective already makes the technology consistent with Trump\u2019s directive.<\/p>\n Microsoft, a major supplier of online services to the government, declined to comment.<\/p>\n Musk\u2019s xAI, through spokesperson Katie Miller, a former Trump official, pointed to a company comment praising Trump\u2019s AI announcements but didn\u2019t address the procurement order. xAI recently announced it was awarded a U.S. defense contract for up to $200 million, just days after Grok publicly posted a barrage of antisemitic commentary that praised Adolf Hitler.<\/p>\n Anthropic, Google, Meta, and Palantir didn\u2019t respond to emailed requests for comment Thursday. <\/p>\n The ideas behind the order have bubbled up for more than a year on the podcasts and social media feeds of Trump\u2019s top AI adviser David Sacks and other influential Silicon Valley venture capitalists, many of whom endorsed Trump\u2019s presidential campaign last year. Their ire centered on Google\u2019s February 2024 release of an AI image-generating tool that produced historically inaccurate images before the tech giant took down and fixed<\/a> the product. <\/p>\n Google later explained that the errors \u2014 including generating portraits of Black, Asian and Native American men when asked to show American Founding Fathers \u2014 were the result of an overcompensation for technology that, left to its own devices, was prone to favoring lighter-skinned people because of pervasive bias in the systems. <\/p>\n Trump allies alleged that Google engineers were hard-coding their own social agenda into the product.<\/p>\n \u201cIt\u2019s 100% intentional,\u201d said prominent venture capitalist and Trump adviser Marc Andreessen on a podcast in December. \u201cThat\u2019s how you get Black George Washington at Google. There\u2019s override in the system that basically says, literally, \u2018Everybody has to be Black.\u2019 Boom. There\u2019s squads, large sets of people, at these companies who determine these policies and write them down and encode them into these systems.\u201d<\/p>\n