Study Finds Grok Leans Republican While OpenAI Favors Democrats in Elections

24 Oct 2026, 8:30 pm · 3h ago · 1 min read · TechCrunch AI

A new research study reveals distinct political biases in leading artificial intelligence models when prompted to evaluate political races. According to the findings, xAI's Grok consistently favored Republican candidates, whereas OpenAI's GPT models predominantly recommended Democrats. The two major AI systems disagreed in more than 90% of nearly 500 election contests analyzed during the study. These findings highlight growing concerns over embedded political leanings in generative AI tools as more users consult them for civic information. The research underscores the challenge developers face in mitigating bias during model training and reinforcement learning processes worldwide.