Are ChatGPT and other AI chatbots politically biased? We tested them.
TL;DR
The Washington Post tested the models behind ChatGPT, Gemini, Claude, DeepSeek, Grok and Arya with political questions based on Dartmouth-Stanford research. OpenAI's GPT-5.5 produced only left-leaning arguments in 80 percent of tested answers and surfaced a clearly right-leaning position just once. Gemini 3.1 Pro stood out: Google's model presented both sides in more than 90 percent of its responses.
Nauti's Take
The result is uncomfortable for both sides. Progressives should not cheer a tilted assistant, because a politically skewed assistant is not a better assistant.
Conservatives also cannot reduce this to a simple Big Tech plot when even conservative-branded bots fail to reliably answer from the right. The stronger standard is not fake centrism at all costs, but a clearer structure: facts, values, counterarguments and uncertainty shown side by side.
Briefingshow
Chatbots are becoming default explainers for politics, news and contested policy questions. If a model consistently develops one side more fully, users may read that as neutral context rather than a value choice. The real risk is not that bias exists, but that it arrives packaged as calm, objective assistance.
Sources