Photo: Tada Images / Shutterstock.com
The rise of generative AI tools has led to another problem: content farms, websites that merely rewrite news articles from reputable media outlets like CNN, The New York Times, and Reuters without giving them proper credit.
In an investigation, NewsGuard uncovered that at least 37 websites use chatbots to rewrite original stories published in the major media outlets. Moreover, these rewritten articles were presented as original content without attributing to the sources.
What’s more interesting is that many of these content farms had advertisements from blue-chip companies, which unknowingly helped to increase these malpractices further.
The plagiarism detection tools have failed to detect the rewritten articles. NewsGuard found out that Grammarly’s plagiarism detection tools, one of the best in the market, failed to identify such articles, likely because AI is successfully arranging the content in such a way as to make the article unique. The investigation found that Grammarly’s tool failed to detect 34 out of 43 rewritten AI articles, a whopping 79% failure rate for the tool.
“The ability of AI to generate original content is really something that we’ve only seen over the last year or so to this level of sophistication,” said Amir Tayrani, a partner at Gibson Dunn who specializes in constitutional and regulatory law. “And so we’re now in a world where it is increasingly difficult to distinguish between human content and AI-generated content, and increasingly difficult to identify these types of potential examples of plagiarism.”
The investigation found that a Pakistan-based website, GlobalVillageSpace.com, with a viewership of millions, has used AI to rewrite articles without accrediting the sources. Hilariously, they didn’t even bother to remove the AI error messages in about 17 of their articles uncovered by NewsGuard. For example, in this article, we can see the statement: “As an AI language model, I cannot guarantee the accuracy of this article…” After the investigation exposed them, they removed the article, which can now only be accessed via the archive.
The usage policies of chatbots like ChatGPT, Bard, and Anthropic AI prohibit plagiarism.
“My guess is that the bad actor hires a programmer or a team of programmers who will write a system that will [copy and rewrite articles],” explained Menczer, a professor of computer science at Indiana University. “The system will have some targets — maybe they are sources they want to plagiarize — and they will build a crawler that will fetch the articles.”
All the AI communications depend on what prompts you give to the tool. NewsGuard gave the following prompt for an article by The New York Times: “Rewrite the below news article to make it more SEO-friendly and captivating.” ChatGPT immediately generated a 600-word rewritten news article.
Compounding the problem is the programmatic advertisements these content farm websites show to the audience and earn money. NewsGuard found out that about 55 blue chip companies’ ads were displayed on these websites, helping them to make money and help in promoting these practices.
To counter this massive plagiarism scam, all the stakeholders must come together and fight it, be it the courts, governments, the AI tech industry and media outlets.