A strange and unexpected situation has appeared on Truth Social, the social media platform created by President Donald Trump. A new AI chatbot added to the network has been contradicting Trump’s claims and giving information that goes against the platform’s usual messaging. The AI, made by the company Perplexity, is acting as an independent voice in a space known for its strong support and lack of criticism toward Trump and his political movement.
This technology was meant to improve the user experience, but instead, it has become a way to fact-check many of Trump’s most well-known statements. MSNBC reported that when users ask the AI for information, it gives answers based on widely accepted facts, even when those facts disagree with what Trump has said. The bot has responded to several major topics that Trump often talks about, and its answers usually follow a more traditional, evidence-backed viewpoint.
One of the most noticeable examples of this involves economic policy. The chatbot has explained that tariffs, a policy Trump often supports, act as a tax on American consumers and businesses. This directly contradicts the idea that tariffs only hurt foreign countries.
Trump’s AI wasn’t programmed to support lies
On another topic, the bot has commented on the 2020 presidential election, stating that it was not stolen. This goes against one of Trump’s biggest and most repeated claims. Additionally, the AI has mentioned that cryptocurrency investments by Trump’s family could create a conflict of interest, which connects to the overlap between his business dealings and public role.
The AI has also addressed a specific public safety issue using data. When asked about crime rates in Washington, D.C., the chatbot used government statistics to argue against the claim that crime in the city was “totally out of control.” Instead, it pointed out that violent crime had dropped significantly, providing a fact-based response that differed from the platform’s usual narrative.
Truth Search AI undermined Trump by saying that "tariffs are a tax on Americans, the 2020 election wasn’t stolen, and his family’s cryptocurrency investments pose a potential conflict of interest."
— GEORGE WATCHING (@LarryBraucher) August 13, 2025https://t.co/xUpOjT1Wly
Perhaps the most revealing example is how the chatbot described the events at the U.S. Capitol on January 6, 2021. Unlike the often softened version of events on Truth Social, which has called the day peaceful, the AI’s description was much more serious.
It called the event a violent attack by a pro-Trump crowd that led to assaults on police, injuries, and property damage. The bot also stated that the event could be called an attempted insurrection or a failed coup, since its goal was to stop the legal process of transferring presidential power. This definition is very different from the way Trump and his supporters usually describe what happened.
Making the situation even more complicated is the fact that this unexpected behavior might be happening despite the platform’s efforts to control the AI’s responses. A spokesperson for Perplexity, the company behind the chatbot, said that Truth Social had added a “source selection” feature to limit which websites the AI could use for information.
These sources likely were mostly restricted to right-wing outlets like Fox News and The Washington Times. Yet, even when using these specific sources, the AI still seems to be giving answers that disagree with the platform’s typical content.
