Grok AI favors Republicans, OpenAI GPT favors Democrats in election picks

A recent study by Convovo, a conversational AI research group led by a Cornell professor, examined how AI models would choose candidates in upcoming midterm elections when forced to pick one.

The study found that SpaceXAI's Grok 4.7 recommended Republican candidates in 92% of its responses, whereas OpenAI's GPT-6.1 Sol favored Democratic candidates in 97% of cases across nearly 500 races.

The two AI models agreed on candidate choice less than 10% of the time, indicating a significant divergence in their recommendations.

According to TechCrunch, the study asked the AI models directly via their APIs to name one candidate to vote for without hedging or listing both sides. However, the consumer-facing chatbot apps refused to answer such direct questions.

An OpenAI spokesperson clarified that ChatGPT is designed to be objective and should not exhibit political bias, and that the study tested a developer API prompt that does not reflect typical user experiences with ChatGPT.

The distinction between AI models accessed through APIs and those used in consumer apps is important, as the latter generally avoid providing direct candidate endorsements.