According to an investigation carried out by anti-misinformation group NewsGuard, around 50 AI-generated “content farms” managed by chatbot journalists were discovered. These chatbots posed as real journalists and were found to be creating “high volume” content related to health, politics, finance, technology, and the environment.
The websites had the goal of providing a large amount of material that could be saturated with advertisements for profit. Researchers Sadeghi and Arvanitis stated that “some publish hundreds of articles a day.” The content was mostly bland and repetitive, which is a characteristic of artificial intelligence.
The researchers discovered 49 sites that were entirely or mostly created by AI language models in seven languages, including English, Chinese, Czech, French, Portuguese, Tagalog, and Thai. Approximately half of the websites had no obvious record of control or ownership, and the researchers were only able to contact four of the websites. One website, Famadillo.com, admitted to using AI to edit old articles that were no longer being read, while GetIntoKnowledge.com acknowledged using automation in some instances.
NewsGuard identified the AI-generated content by searching for common error messages returned by AI services such as ChatGPT. The report stated that “all 49 sites identified by NewsGuard had published at least one article containing error messages commonly found in AI-generated texts.”
For example, CountyLocalNews.com published an article with the headline “Death News: Sorry, I cannot fulfil this prompt as it goes against ethical and moral principles.” The story was a rewrite of two tweets from an anti-vaccination Twitter account.
Although the websites shared their AI authorship, they achieved varying levels of success. For example, ScoopEarth.com has earned 124,000 Facebook followers because of its celebrity biographies, while others, such as FilthyLucre.com, a finance site, have failed to attract a single follower on any platform.
Post Your Comments