TrustLens-AI: A Multidimensional Analytical Framework for Measuring Public Trust in AI-Generated News

Abstract-This study presents TrustLens-AI, a multidimensional framework for quantifying public trust in AI-generated news, combining perception metrics with sentiment and contextual adjustments. Using survey data from 62 participants, we calculated a composite TrustLens Score (TLS) from five weighted dimensions: content quality, transparency cues, ethical alignment, engagement sentiment, and prior AI exposure, adjusted for platform reputation and consumption frequency. The mean TLS was 64.3 ± 12.7 (0–100 scale). Disclosure of AI authorship increased mean trust ratings from 3.1 to 3.8 (1–5 scale), while respondents with prior AI news exposure scored 9.4 points higher in TLS on average. Decision tree analysis (R² = 0.74, MAE = 0.42) identified transparency cues and content quality as the strongest predictors, followed by ethical alignment. These results highlight that targeted transparency strategies and ethical oversight can substantially improve perceived credibility, helping bridge the trust gap between AI-generated and human-authored journalism.摘要：本研究提出TrustLens-AI——一款用于量化公众对人工智能生成新闻信任度的多维分析框架，该框架融合感知指标与情感、上下文校正机制。研究采用62名受访者的调研数据，从内容质量、透明度提示、伦理契合度、参与情感及既往人工智能接触度五个加权维度，计算得到综合信任度评分（TrustLens Score，简称TLS），并针对平台声誉与内容消费频率进行了校准。该综合评分的均值为64.3±12.7，评分区间为0~100。披露人工智能内容的作者身份，可使平均信任评级从3.1提升至3.8（采用1~5分制）；既往接触过人工智能生成新闻的受访者，其TLS平均得分较其他受访者高出9.4分。决策树分析结果显示（决定系数R²=0.74，平均绝对误差MAE=0.42），透明度提示与内容质量为最强的预测因子，伦理契合度紧随其后。本研究结果表明，针对性的透明度优化策略与伦理监管机制，可显著提升人工智能生成新闻的感知可信度，有助于弥合人工智能生成新闻与人类原创新闻之间的信任鸿沟。