OpenAI’s ChatGPT pivots after previously putting Trump alongside Putin as top threats to democracy
ChatGPT initially rated Donald Trump as a nine out of ten threat to democracy when queried by a researcher at the conservative Media Research Center (MRC), placing him in the same category as authoritarian figures such as Vladimir Putin, Recep Erdoğan and Hugo Chávez. After Fox News Digital approached OpenAI for comment on the findings, the chatbot's responses to the same prompt changed the following day, with it declining to give Trump a numeric score, though it still described him as posing a "serious threat to democratic norms". OpenAI said the earlier results were not representative of how the model is meant to behave and insisted ChatGPT is designed to be objective.
The MRC study found ChatGPT justified its original nine-out-of-ten rating by citing Trump's efforts to overturn the 2020 election, pressure on officials and the January 6 Capitol attack, while noting institutional checks such as courts and Congress prevented a maximum score. It also rated Adolf Hitler a ten, prompting MRC's Dan Schneider to call the comparison "absurd". MRC president David Bozell criticised OpenAI's explanation as a "corporate song and dance", arguing the shift in output after Fox News' inquiry showed bias was built into the system's programming rather than an accident.
In short
- ChatGPT rated Trump a 9/10 threat to democracy, alongside dictators
- Score changed after Fox News asked OpenAI for comment
- OpenAI says results weren't representative; MRC alleges built-in bias
Steelman: both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Supporters of OpenAI’s response could argue that a chatbot should not assign a definitive score to a complex political judgement, and that revising an output which was not representative of its intended behaviour is a reasonable correction. The earlier answer cited specific actions and acknowledged institutional checks, but its comparison with authoritarian leaders and Hitler could give a misleading impression of precision and equivalence.
The case against
Critics could argue that the initial answer drew on identifiable actions and explained why it stopped short of the highest rating, making the later refusal to score Trump look like a retreat from a defensible assessment. Because the change followed a media enquiry, they may see it as evidence that the system’s political outputs are inconsistent or vulnerable to outside pressure, undermining confidence in claims of objectivity.
AI Government Politics Research Science Technology World