OpenAI Claims GPT-5 Shows 30% Less Political Bias Than Previous Models in New Testing

OpenAI reports GPT-5 shows 30% less political bias than previous models after testing 500 questions across 100 topics with conservative and liberal prompts, promising more transparency results soon.

• OpenAI claims GPT-5 demonstrates 30% reduced political bias compared to previous models in both instant and thinking modes
• Researchers tested the model using 500 questions across 100 topics with prompts ranging from conservative charged to liberal charged scenarios
• The company plans to publish additional bias testing results in coming months to promote transparency and help other AI companies build similar evaluations