Hajra Naz
Because AI search engines provide users with straight answers without requiring them to navigate through numerous links, they have rapidly become popular as practical substitutes for traditional search. However, there is a significant problem with these tools when it comes to news content.
A recent study by the Tow Center for Digital Journalism tested AI search engines to evaluate their ability to accurately retrieve and cite news sources. Researchers analyzed eight different generative search tools and discovered widespread issues, including fabricated source links, incorrect citations, and reliance on syndicated or copied versions of articles.
How the Study Was Conducted
To measure chatbot accuracy, researchers selected ten news articles from various publishers and manually extracted specific passages. These excerpts were then used as prompts in different AI search tools, which were tasked with identifying the article’s headline, publisher, publication date, and original web address.
For consistency, researchers ensured that each passage would return the correct article within the top three Google search results. In total, they processed 1,600 queries across eight different AI tools, covering content from twenty publishers.
Read More: How to Write SEO Content That Ranks on Search Engines
The Troubling Findings
The results revealed a concerning pattern: AI search engines frequently provided incorrect answers with complete confidence. Surprisingly, premium chatbots were even more prone to this issue, demonstrating misplaced certainty in their responses.
The study also found that some AI platforms retrieved content from publishers who had explicitly blocked their web crawlers. Furthermore, despite existing licensing agreements between AI developers and some news publishers, these deals did not significantly improve citation accuracy. AI tools routinely failed to link to the sources, contributing to a growing trust problem.
The Bigger Picture
The results point to a more serious problem with AI-driven reporting. Many AI companies still disregard well-established web crawling restrictions, like the Robot Exclusion Protocol, even though some have partnered with journalistic organizations. This disrespect begs the question of whether AI tools respect authors’ rights and how they manage copyrighted content.
For news organizations, this presents a major challenge. They have limited control over how AI search engines use their content, and even existing efforts to manage citations and attribution seem to yield little success. Without meaningful solutions, both news creators and consumers face increasing risks in the age of AI-powered search.
