We study how AI-generated misinformation affects demand for trustworthy news, using data from a field experiment by Süddeutsche Zeitung (SZ), a major German newspaper. Readers were randomly assigned to a treatment that highlighted the difficulty of distinguishing real from AI-generated images. The treatment increased concern over misinformation (+0.3 s.d.) and reduced trust in all news sources (-0.05 to -0.1 s.d.), including SZ itself. Crucially, it also affected post-survey browsing behavior: daily visits to SZ digital content rose by 2.5% in the days following the treatment. In addition, subscriber retention increased by 1.2% over the following five months, corresponding to about a one-third reduction in the attrition rate. These results are consistent with a model in which the relative value of trustworthy news sources rises with the prevalence of misinformation, boosting engagement with these sources even as trust in news content declines.

• Exposição a IA faz engajamento com mídia tradicional subir, indica estudo Folha de S. Paulo, Aug 07, 2025
• Trusted news sites may benefit in an internet full of AI-generated fakes, a new study finds Nieman Lab, Aug 28, 2025
• At least five interesting things: Cool research edition (#68) Noahpinion, Aug 12, 2025
• Rise of AI-Generated Content Could Help News Sites, Research Reveals Futureweek, Sep 01, 2025