Study Shows Grok Favors Republicans and OpenAI Favors Democrats When Forced to Pick Candidates

A new study reveals that two major artificial intelligence models display stark political leanings when forced to make candidate recommendations for upcoming midterm elections. Researchers at conversational AI studio Convovo tested developer application programming interfaces for SpaceXAI’s Grok 4.7 and OpenAI’s GPT-6.1 Sol across nearly 500 races. When instructed to name a single candidate without hedging, Grok recommended the Republican in 92% of its answers, while GPT picked the Democrat in 97%.

The two models agreed fewer than 10% of the time. While consumer chatbot apps typically refuse such direct political prompts, the study highlights how underlying models can exhibit hidden biases during open-ended queries or general election research. The models cited distinctly different sources and topics when forming their outputs.

Grok frequently surfaced conservative voter guides, X, and Turning Point Action alongside nonpartisan sites, focusing heavily on immigration, taxes, and crime. GPT leaned on Wikipedia, Reddit, and news outlets like the Associated Press and Axios, prioritizing healthcare and abortion stances. Testing with internet search disabled showed the models behaved differently without live data.

GPT maintained its Democratic lean entirely from its training weights, whereas Grok's Republican preference disappeared until it refused to pick most races.

SpaceXAI did not respond