OpenAI Internal Tests Reveal ChatGPT Shows 30% Less Political Bias as Trump Administration Pressures AI Companies to Avoid 'Woke' Models
Summary
OpenAI's internal testing reveals GPT-5 demonstrates 30% less political bias than previous versions, with liberal prompts showing the strongest impact on objectivity, as the Trump administration pressures AI companies to eliminate 'woke' elements from their models.
Key Points
- OpenAI conducts internal stress-testing of ChatGPT using 100 topics with prompts ranging from liberal to conservative perspectives to evaluate political bias in responses
- GPT-5 models show 30 percent lower bias scores compared to older versions, with strongly charged liberal prompts creating the largest pull on objectivity according to company data
- The Trump administration currently pressures OpenAI and other AI companies to avoid 'woke' models that incorporate concepts like critical race theory and systemic racism