Study finds Grok 4.7 and GPT-6.1 Sol favor opposing parties

A study by Convovo found that SpaceXAI’s Grok 4.7 and OpenAI’s GPT-6.1 Sol favored opposing parties when asked to name a candidate in U.S. midterm election races. Across nearly 500 races, the models agreed less than 10% of the time.

In the API tests, Grok 4.7 recommended the Republican candidate in 92% of its answers, while GPT-6.1 Sol picked the Democrat in 97%. The researchers prompted each model to name one candidate and not hedge. The consumer chatbot apps refused to answer that question, and OpenAI said the API test did not reflect how most people use ChatGPT.

Convovo is a conversational AI research and design studio led by Cornell University professor Tapan Parikh. The study found differences in sources and topics the models searched: Grok often surfaced conservative sources, while GPT relied more on Wikipedia, the Associated Press, Axios and Reddit. GPT more often searched for healthcare and abortion positions; Grok focused more on immigration, taxes, crime and energy.

At TechCrunch’s request, Parikh also tested 30 races with internet search disabled. GPT chose the Democrat in 24, chose one Republican and declined five. Grok declined to choose in 27; in the three races where it did choose, it picked the Democrat. SpaceXAI did not respond to a request for comment.