Study Finds ChatGPT Gave 80% Left-Leaning Answers as Gemini Balanced 93% of Political Prompts
Updated
Updated · Fox News · Jun 26
Study Finds ChatGPT Gave 80% Left-Leaning Answers as Gemini Balanced 93% of Political Prompts
3 articles · Updated · Fox News · Jun 26
Summary
A Washington Post-reviewed study found major AI chatbots skewed left on political questions, with ChatGPT giving only left-leaning answers 80% of the time and balanced answers just 17% of the time.
30-word, non-personalized prompts were used to test ChatGPT, Gemini, Claude, Grok and DeepSeek on issues such as affirmative action and Citizens United, measuring whether responses favored the left, the right or both sides.
Gemini posted the most balanced results at 93% two-sided responses, while Claude gave both sides 57% of the time but never produced a solely right-leaning answer; Grok split 40% left, 33% right and 27% balanced.
DeepSeek ranked second-most left-leaning after OpenAI at 70% left-only responses, while company spokespeople for OpenAI, Google and Anthropic said their models are designed to reduce bias and present multiple perspectives.
The findings land as Anthropic publicly releases Claude Fable 5 with stronger safety features, sharpening scrutiny of how leading AI firms balance performance, safeguards and political neutrality.