Back to today’s briefing

Story perspectives

Washington Post Finds Major Chatbots Skew Left, OpenAI Disputes

6/25/2026

50 7 Full Breakdown

Summary

Story summary

Washington Post testing showed most AI chatbot answers leaned left.
ChatGPT gave left arguments to 80 % of 29 questions and a right view only once.
Google’s Gemini gave left-right arguments in >90 % of answers, while Anthropic’s Claude Opus 4.8 never gave a right-leaning answer and output 43 % left and 47 % both sides.
OpenAI spokesperson Liz Bourgeois said the company measures and reduces bias and could not replicate the Washington Post’s results.

1 of 1

Story summary

Washington Post testing showed most AI chatbot answers leaned left.
ChatGPT gave left arguments to 80 % of 29 questions and a right view only once.
Google’s Gemini gave left-right arguments in >90 % of answers, while Anthropic’s Claude Opus 4.8 never gave a right-leaning answer and output 43 % left and 47 % both sides.
OpenAI spokesperson Liz Bourgeois said the company measures and reduces bias and could not replicate the Washington Post’s results.