As AI increasingly influences content production in news, and at a time when media organizations aim to restore credibility through higher transparency, public perceptions of AI play a role in shaping evaluations of the message and the source. This between-subject experiment ( n = 415) examined how the label used to describe AI involvement (“influenced,” “assisted,” “generated”), the media format (blog vs news agency), and individual differences in perceptions of AI’s editorial value and AI literacy influence credibility. Results indicated that message credibility varied significantly by label, with AI-generated content perceived as least credible and AI-influenced content as most credible. This pattern may reflect audiences’ preference for human judgment in content creation and greater skepticism toward messages produced entirely by AI. In contrast, source credibility was unaffected by the label but was higher for a news agency than for blogs, likely due to trust in perceived professional standards associated with institutional journalism. Perceived AI editorial value emerged as a strong predictor of source credibility, whereas AI literacy did not moderate these effects. Findings suggest that while audiences negatively respond to full AI authorship in evaluating message credibility, trust in the source depends more on editorial value than AI labeling.
Merle et al. (Thu,) studied this question.