Fact-Checking Misinformation in the Age of AI: Audience Responses to Human, AI, and Hybrid Sources

As artificial intelligence (AI) increasingly participates in fact-checking, it remains unclear how audiences evaluate AI fact-checking compared to human fact-checking. Grounded in machine heuristic theory, this study used a 3 (source: human vs. AI vs. human–AI collaboration) × 2 (topic: health vs. society) × 2 (correction method: fact-based vs. logic-based) online experiment (N = 480) to test how source, topic, and correction method shape machine heuristic, perceived quality, and attitudes. Results showed that AI and human–AI fact-checking evoked stronger machine heuristic than human fact-checking. However, AI fact-checking was associated with lower perceived quality and more negative attitudes, while human–AI collaboration increased perceived quality. Social topics unexpectedly heightened machine heuristic, and machine heuristic mediated AI’s source effects on perceived quality and attitudes. Algorithm appreciation moderated the effect of AI fact-checking on machine heuristic and the effect of machine heuristic on perceived quality. These findings extend understanding of audience responses to AI fact-checking.