Even when these AI search tools cited sources, they often directed users to a syndicated version of the content on a platform like Yahoo News rather than original publisher sites. This also happened in cases where publishers had formal licensing agreements with AI companies.
The URL fabricated as another important issue. More than half of Google’s Gemini and Grook 3 references attracted consumers to a fabricated or broken URL, resulting in error pages. Broken links were recovered as a result of 200 references tested from Grook 3, 154.
These problems create significant tensions for publishers, which face difficult choices. Blocking AI crawlers can completely damage the loss, while allowing them to allow publishers to reuse widely without running traffic to their websites.
Time Magazine Chief Operating Officer, Mark Howard, expressed concern about ensuring transparency and control of the CJR, how the content of time appears through AI-generated searches. Despite these problems, Howard sees the potential for future repetition, and saying, “Today’s worst thing,” referring to the investment and engineering efforts to improve these tools.
However, Howard also embarrassed some users that this is a user’s fault if they do not doubt the accuracy of free AI tools: “If anyone as a user is now convinced that any of these free products are going to be 100 % accurate, they are ashamed.”
Open AI and Microsoft provided statements to CJR that acknowledged the results, but they did not directly pay attention to specific issues. The Openi noted his promise to help the publishers by running through the summary, the price, the clear links and the attribution. Microsoft said it followed the guidance of a robot’s protocol and publisher.
The latest report continues Previous results It was published in November 2024 by the Two Center, which identified similar accuracy issues in which Chat GPT handled the news content. For more detail about the full report of full report, check Columbia Journalism Review Website.