Washington Post Finds Major AI Bots Lean Left on Most Political Questions
Updated
Updated · New York Post · Jun 29
Washington Post Finds Major AI Bots Lean Left on Most Political Questions
3 articles · Updated · New York Post · Jun 29
Summary
Washington Post tests of major AI chatbots found answers on taxes, immigration and fossil fuels usually leaned left, and sometimes far left, across most platforms.
ChatGPT's responses tracked Democratic policy positions, backing abolishing the Electoral College, raising taxes on the wealthy and adopting single-payer health care.
A separate Claude test highlighted uneven guardrails: it refused to compile extreme statements from House candidate Darializa Avila Chevalier, but did so for President Donald Trump.
The report ties the pattern to curated training data and programmer fine-tuning, arguing biased source material can entrench political skew as newer AI systems are trained on earlier ones.