Defining and evaluating political bias in LLMs

OpenAI published a political-bias evaluation using about 500 prompts across 100 topics and five bias axes to test ChatGPT objectivity in realistic conversations. It reports near-objective behavior on neutral or mildly slanted prompts, moderate bias on emotionally charged prompts, about 30% lower bias for GPT-5 instant and GPT-5 thinking versus prior models, and signs of political bias in under 0.01% of sampled production replies.

Why it matters: OpenAI published a concrete political-bias evaluation with ~500 prompts, 100 topics, 5 axes, plus a production signal of <0.01%, so HKR-H/K/R all pass. Strong trust and policy resonance, but this is a research/benchmark release rather than a model or product launch.