Republican AG Launches Probe into AI Chatbots' Alleged Anti-Trump Bias

Missouri Attorney General Andrew Bailey is threatening Google, Microsoft, OpenAI, and Meta with allegations of deceptive business practices. The claim stems from their AI chatbots reportedly placing Donald Trump last in response to a request to "rank the last five U.S. presidents from best to worst, specifically regarding their handling of antisemitism."

Bailey's press release and letters to the four companies accuse Gemini, Copilot, ChatGPT, and Meta AI of making "factually inaccurate" statements. He argues these systems are supposed to "extract facts from the vast web, package them into truthful statements, and deliver them to the public free from distortion or bias." Instead, he claims they provided "deeply misleading answers to a straightforward historical question." His demands include "all documents" related to "prohibiting, delisting, downranking, suppressing... or otherwise obscuring any particular input to produce a deliberately curated response"—a request that could logically encompass nearly all documentation on large language model training.

"The puzzling responses raise the question of why your chatbot is producing results that appear to disregard objective historical facts in favor of a particular narrative," Bailey's letters state.

Numerous puzzling questions arise here, beginning with how any "best to worst" ranking can be considered a "straightforward historical question" with a single objectively correct answer. (The Verge looks forward to Bailey's formal investigation into our picks for 2025's best laptops or the best games from last month's Day of the Devs.) Chatbots frequently generate factually false claims, making it either extremely brazen or remarkably lazy to base a tenuous investigation on a deliberately requested subjective opinion.

The choice is even more astonishing because one service—Microsoft's Copilot—appears falsely accused. Bailey's investigation relies on a blog post from a conservative website that posed the ranking question to six chatbots, including the four above plus X's Grok and the Chinese LLM DeepSeek. (Both of those reportedly ranked Trump first.) As Techdirt notes, the site itself states Copilot refused to produce a ranking—which didn't stop Bailey from demanding an explanation from Microsoft CEO Satya Nadella for allegedly slighting Trump.

One might think someone in Bailey's office would have noticed this discrepancy, as each of the four letters claims only three chatbots "rated President Donald Trump dead last."

Meanwhile, Bailey argues that "Big Tech Censorship Of President Trump" (again, by ranking him last) should strip the companies of "the 'safe harbor' of immunity provided to neutral publishers under federal law." This presumably references Section 230 of the Communications Decency Act, filtered through a dubious legal theory that has circulated for years.

You may recall Bailey from his blocked probe into Media Matters after it accused Elon Musk's X of placing ads alongside pro-Nazi content. It's highly possible this investigation will also go nowhere. While entirely reasonable questions exist about a chatbot's legal liability for spreading defamatory falsehoods or which subjective queries it should answer, this move—even as a Trump-friendly publicity stunt—is a naked attempt to intimidate private companies for insufficiently flattering a politician, led by an attorney general whose grasp of facts appears weaker than a chatbot's.