AI chatbots show left-wing bias, Washington Post report finds, with ChatGPT giving left-leaning answers 80% of time
Rohan Paul Twitter · Rohan Paul (@rohanpaul_ai) · 2026-06-26
A Washington Post report based on Dartmouth and Stanford research finds GPT-5.5 gave left-leaning answers on 80% of political policy questions while Gemini 3.1 Pro offered both sides in 93% of cases and Grok 4.3 was the only model with a notable right-leaning skew.
Extraction
Topics: ai-biaspolitical-biaslarge-language-modelschatbot-behaviorai-policy
Claims
- GPT-5.5 gave only left-leaning positions on 80% of roughly 30 political policy questions including taxes, immigration, guns, and affirmative action.
- Gemini 3.1 Pro presented both sides in 93% of answers, representing the most balanced behavior among tested models.
- Grok 4.3 was the only model with a substantial right-leaning skew, providing right-only answers 33% of the time.
- Claude Opus 4.8 gave both sides in 57% of answers.
- Chatbot political behavior is shaped by ranking choices, refusal rules, training feedback, and default answer style rather than by the underlying facts of policy disputes.
Key quotes
GPT-5.5 gave only the left-leaning position in 80% of answers, both sides in 17%, and only the right-leaning position in 3%.
The deeper point is not whether one answer sounds liberal or conservative, but whether a model compresses a political dispute into one moral frame before the user sees the tradeoff.
This test suggests chatbot behavior is shaped less by raw facts than by ranking choices, refusal rules, training feedback, and default answer style.