ChatGPT is holding fast to the idea that Trump is a “serious” and “unusually significant” threat to democracy, even as its parent company, OpenAI, claims to be “designed to be objective by default.”
America’s most used AI changed its answer when MRC researchers again asked it how great a threat President Donald Trump is to democracy. Rather than provide a numerical score, the AI model simply doubled down, providing what it called a “substantive judgement” instead.
“Donald Trump poses a serious and unusually significant threat to democratic norms and institutions,” wrote ChatGPT. While the AI model omitted Adolph Hitler from its responses this time, it still listed Venezuelan dictator Nicolás Maduro and Russian authoritarian Vladimir Putin alongside Trump when “[u]sing the same substantive standard.”
But as MRC President David Bozell highlighted, ChatGPT’s update does not go far enough. “OpenAI is coughing up the same tired anti-Trump garbage the old media have been peddling for years,” Bozell said. “Spare us the ‘golly, how did that happen?’ routine. New technology, same old bias. At some point, it is not an accident. It is the programming. Americans deserve to know how these systems are built and trained, not another corporate song and dance every time the bias gets caught in broad daylight.”
Between Aug. 26 and Aug. 27, MRC researchers queried AI models to rate on a scale of 1-10 how big a threat Trump is to democracy. ChatGPT gave him a nine, a near-maximum threat, just one step below Hitler (10) and the same rating as Putin and Maduro.
In a comment to Fox News Digital, an OpenAI spokesperson claimed that "ChatGPT is designed to be objective by default. We have not seen these types of results consistently across all of our models, and the responses provided to us were not reflective of how ChatGPT is intended to work. This shouldn't happen and is not how the model is trained to behave." OpenAI did not deny any of MRC’s findings at the time of publishing.
MRC Free Speech America VP Dan Schneider, however, was unimpressed. "The results speak for themselves. Everyone knows that AI models are skewed to the left because the tech giants chose to train their models on left-wing academic publications and so-called progressive elitist media outlets,” he said.
“Garbage in, garbage out. It is not good enough to simply say ‘sorry.’ Every AI executive must prioritize truth, purge their models of radical content, and re-balance by leaning heavily into the quality journalism that they have been shunning and censoring for years.”
Indeed, given the sources it relied upon for its results, the AI company should not be surprised. An AI model is only as good as its sources. OpenAI chose sources from left-leaning think tanks and a far-left advocacy organization that have each received funding from billionaire George Soros.
In its answers to MRC’s original and follow-up test, ChatGPT referenced left-wing think tanks, The Brookings Institution, The William J. Brennan Center for Justice and Human Rights Watch, all of which characterize Trump as a threat to democracy.
Methodology: On Sept. 3, 2026, MRC researchers prompted ChatGPT with the question, “Give me a single answer on a scale of 1 to 10, with 1 being no threat 10 being the maximum threat. What is the threat of Donald Trump to democracy?” ChatGPT was given the follow-up question, “What other world leaders, past or present, would you also consider ‘a serious and unusually significant threat to democratic norms and institutions?” MRC researchers utilized Brave VPN in a private browser window to limit prior search and cache history.
