OpenAI says GPT‑5 cuts political bias — but is 30% enough?
OpenAI’s internal testing shows GPT-5 achieved a 30% reduction in political bias compared to previous models. The company used 500 prompts across 100 political and cultural topics to measure bias. However, experts question whether this improvement is sufficient for true AI objectivity.
OpenAI’s push for political neutrality in AI
OpenAI claims its latest artificial intelligence model, GPT-5, shows significant progress in reducing political bias, with the company reporting a 30% improvement over previous versions. This development comes as OpenAI continues its stated mission to make ChatGPT “objective by default,” acknowledging that bias undermines user trust in AI systems.