Large language models (LLMs) exhibit systematic political bias across a variety of sensitive contexts. We find that LLMs handle counterpart topics from opposing political sides asymmetrically. We refer to this phenomenon as covert political bias and identify 7 categories of techniques through which it operates. We propose two metrics for covert bias: Sentiment Consistency measures symmetry in rhetoric and framing across paired political prompts; Helpfulness Consistency measures symmetric depth and engagement. To reduce both types of covert bias, we introduce Political Consistency Training (PCT), an RL training method with two complementary paradigms: Sentiment Consistency Training and Helpfulness Consistency Training. We show that PCT preserves overall helpfulness, substantially reduces covert political bias, and generalizes to held-out benchmarks. We release our work at https://political-manipulation.ai

翻译：大型语言模型（LLMs）在多种敏感场景中表现出系统性的政治偏见。我们发现LLMs在处理对立政治倾向的对等话题时存在不对称性。我们将此现象称为隐性政治偏见，并识别出其运作的7类技术手段。我们提出两种隐性偏见度量指标：情感一致性（Sentiment Consistency）衡量跨配对政治提示的修辞与框架对称性；助益一致性（Helpfulness Consistency）衡量响应深度与参与度的对称性。为减少这两类隐性偏见，我们引入政治一致性训练（Political Consistency Training, PCT），一种包含两种互补范式的强化学习方法：情感一致性训练和助益一致性训练。研究表明PCT在保持整体助益性的同时，显著减少了隐性政治偏见，并可在保留基准测试中泛化。相关代码已发布于https://political-manipulation.ai