Study Finds ChatGPT Misattributes and Misrepresents Publisher Content
1 years ago
A study by the Tow Center for Digital Journalism highlights concerns about ChatGPT's accuracy in citing sources. Despite licensing deals, ChatGPT often misrepresents or invents information, risking publishers' trustworthiness and brand safety. The research found ChatGPT frequently provided incorrect citations, even for those allowing content crawling. This inconsistency and lack of transparency could harm publishers' reputations.
