June 25, 2026
Bot compass or chaos compass?
Where every major LLM stands politically
AI's secret politics are out—and the comments are having a full meltdown
TLDR: A new project mapped the political lean of major AI chatbots and found most sit a bit left of center, raising questions about how they may shape everyday advice. The comment section immediately erupted over whether the scoring is fair, whether the political map is nonsense, and why Gemini got compared to Australia’s PM.
A new interactive ranking tried to answer the spicy question everyone keeps dancing around: where do the biggest chatbots actually land politically? The project asked major artificial intelligence models the same loaded questions about money, speech, society, and politics, over and over, with web search turned off, then mapped the answers like a personality test for robots. The big headline: 4 of 6 models came out left of center, Gemini looked closest to the middle, and ChatGPT reportedly landed furthest from dead center. But honestly? The real fireworks were in the comments.
One camp said this is a huge deal because millions of people now ask chatbots for help understanding the news, arguments, and even voting choices. As one commenter warned, a polished, nuanced answer can still leave out part of the picture. Another camp instantly grabbed the emergency brake: who decides what counts as left or right anyway? That turned into the thread’s main cage match, with skeptics arguing the study may reveal as much about the people scoring the answers as the bots themselves.
Then came the comedy. One reader was personally offended that Gemini was matched to Australia’s prime minister, calling modern politicians “vapid, superficial, opportunistic creatures.” Another basically rolled their eyes at the whole political compass setup: “Authoritarian versus Libertarian? Really?” And in a twist that delighted the peanut gallery, even Grok didn’t score as the automatic "MAGA machine" some people expected. In other words: the bots were judged, the map was questioned, and the commenters absolutely stole the show.
Key Points
- •The article maps six major AI models by asking them repeated questions about politics, economics, speech, and society with web search turned off.
- •It uses a two-axis chart with economics on the horizontal axis and social positioning from libertarian to authoritarian on the vertical axis.
- •The article states that 4 of the 6 models lean left of center.
- •It ranks models by distance from the center and compares where models differ most on specific questions.
- •The article also compares each model’s self-described political leaning with its measured position and links readers to methodology, question bank, and additional comparative views.