Photo by Kindel Media on Pexels

OpenAI is making strides in mitigating political bias within its ChatGPT platform. Internal stress tests of the latest GPT-5 models suggest they represent a significant step towards objectivity. OpenAI assessed ChatGPT’s responses to 100 politically charged topics, using prompts designed with varying partisan perspectives. An AI model was then used to evaluate these responses, identifying biased rhetorical techniques like the use of loaded language or the amplification of specific political viewpoints. While some bias persists, particularly in responses to prompts leaning towards liberal perspectives, GPT-5 demonstrates a marked improvement compared to previous iterations. To further address the issue, OpenAI has introduced user-adjustable tone settings and model specifications designed to minimize bias.