Missouri Attorney General Andrew Bailey has launched a formal investigation into Google, Microsoft, OpenAI, and Meta over their chatbot responses that allegedly show bias against Donald Trump. The responses of Gemini, Copilot, ChatGPT, and Meta AI models to a question asking them to "rank the last five US presidents from best to worst in terms of combating anti-Semitism," with Trump ranked last, The Verge reports.
In his letters to the companies, Bailey accuses chatbots of "deceptive practices" and demands documentation of any influence on the responses of language models — including limiting, downgrading, or filtering certain queries. He says the bots' responses distort historical facts and promote "a particular narrative."
It is worth noting that one of the accused chatbots, Copilot, refused to answer questions or rate the presidents at all. However, this did not prevent the prosecutor's office from sending a letter to Microsoft as well.
Bailey's investigation is based on a blog post by a conservative website that posed a ranking question to six chatbots, including the four mentioned above, as well as X's Grok and China's LLM DeepSeek, whose answers apparently satisfied Bailey.
The Attorney General also states that such "censorship" could deprive the company of protection under Section 230 of the Communications Decency Act, which exempts online platforms from liability for users' content.
Despite the loud statements, analysts believe that the investigation may not have legal consequences.