Sign up to save tools and stay up to date with the latest in AI
bg
bg
1

ChatGPT is hallucinating fake links to its news partners’ biggest investigations

Jun 28, 2024 - niemanlab.org
Nieman Lab's tests reveal that OpenAI's ChatGPT is directing users to broken URLs for at least 10 publications with which OpenAI has licensing deals. These include The Associated Press, The Wall Street Journal, the Financial Times, The Times (UK), Le Monde, El País, The Atlantic, The Verge, Vox, and Politico. Despite the licensing deals stating that ChatGPT will produce attributed summaries of each media company’s reporting and link to their publications’ websites, the AI is generating fake URLs and directing users to 404 errors instead of real article pages.

OpenAI has responded by stating that the citation features promised in its licensing contracts have not yet been launched. The company is working with its news publisher partners to build an experience that blends conversational capabilities with their latest news content, ensuring proper attribution and linking to source material. However, it is unclear how OpenAI can guarantee these features while the underlying ChatGPT product is regularly outputting broken links to those same websites.

Key takeaways:

  • Nieman Lab's tests reveal that OpenAI's ChatGPT is directing users to broken URLs for at least 10 publications with OpenAI licensing deals, including The Associated Press, The Wall Street Journal, and The Atlantic.
  • Despite the licensing deals stating that ChatGPT will produce attributed summaries and link to the media companies' websites, the chatbot is generating fake URLs, leading users to 404 errors instead of actual articles.
  • OpenAI has stated that it has not yet launched the citation features promised in its licensing contracts and is working with news publisher partners to develop an enhanced experience that ensures proper attribution and linking to source material.
  • ChatGPT's hallucinated links often follow the standard URL format of a given website but get the specific words or numbers in that URL wrong, indicating that it is predicting the most likely URL for a story without knowing the actual URL.
View Full Article

Comments (0)

Be the first to comment!