Editor’s Note: Since publication of this report, NewsGuard has identified hundreds of additional unreliable, AI-generated websites. Visit our AI Tracking Center for the latest total number and for our latest coverage.

By McKenzie Sadeghi and Lorenzo Arvanitis | Published on May 1, 2023

Artificial intelligence tools are now being used to populate so-called content farms, referring to low-quality websites around the world that churn out vast amounts of clickbait articles to optimize advertising revenue, NewsGuard found. 

In April 2023, NewsGuard identified 49 websites spanning seven languages — Chinese, Czech, English, French, Portuguese, Tagalog, and Thai — that appear to be entirely or mostly generated by artificial intelligence language models designed to mimic human communication — here in the form of what appear to be typical news websites. 

The websites, which often fail to disclose ownership or control, produce a high volume of content related to a variety of topics, including politics, health, entertainment, finance, and technology. Some publish hundreds of articles a day. Some of the content advances false narratives. Nearly all of the content features bland language and repetitive phrases, hallmarks of artificial intelligence.

Many of the sites are saturated with advertisements, indicating that they were likely designed to generate revenue from programmatic ads — ads that are placed algorithmically across the web and that finance much of the world’s media — just as the internet’s first generation of content farms, operated by humans, were built to do.

In short, as numerous and more powerful AI tools have been unveiled and made available to the public in recent months, concerns that they could be used to conjure up entire news organizations  — once the subject of speculation by media scholars — have now become a reality.

In April 2023, NewsGuard sent emails to the 29 sites in the analysis that listed contact information, and two confirmed that they have used AI. Of the remaining 27 sites, two did not address NewsGuard’s questions, while eight provided invalid email addresses, and 17 did not respond.

NewsGuard exchanged a series of emails, some of which were hard to comprehend, with the self-described owner of Famadillo.com, a site that has published numerous AI-generated product reviews attributed to “admin.” This person, who identified themselves as Maria Spanadoris, denied that the site used AI in a widespread manner. “We did an expert [sic] to use AI to edit old articles that nobody read anymore [sic]just to see how it works,” Spanadoris — who declined a phone call with NewsGuard — said, without elaborating.

Adesh Ingale, who identified himself as the founder of GetIntoKnowledge.com, a site that NewsGuard found to have published AI-generated clickbait articles about history, science, and other topics, responded, “We use automation at some points where they are extremely needed. And yes they are 100% facts checked [sic] so that no false information is created… As a world [sic] is growing towards digital and automation era we have introduced some automation softwares in our work but the results getting out of it are 100% original and regional facts based [sic].” When asked by NewsGuard, Ingale did not elaborate on the site’s use of AI, and claimed that the site’s content is “published manually under human supervision.” Ingale added, “We are the new age of providing knowledge to each and every corner.”

Fake Content Creators and Algorithmically Generated Pages

The 49 AI-driven sites that NewsGuard identified typically have benign and generic names suggesting they are operated by established publishers, such as Biz Breaking News, News Live 79, Daily Business Post, and Market News Reports. 

The AI-generated articles often consist of content summarized or rewritten from other sources. For example, BestBudgetUSA.com, a site that does not provide information about its ownership and was anonymously registered in May 2022, appears primarily to summarize or rewrite articles from CNN. 

The articles themselves often give away the fact that they were AI produced. For example, dozens of articles on BestBudgetUSA.com contain phrases of the kind often produced by generative AI in response to prompts such as, “I am not capable of producing 1500 words… However, I can provide you with a summary of the article,” which it then does, followed by a link to the original CNN report.

The presence of these sorts of phrases is also evidence that these sites likely operate with little to no human oversight.

Many of the AI-generated articles identified by NewsGuard are credited to “Admin” and “Editor,” or have no bylines at all. Other sites feature fake author profiles. For example, HarmonyHustle.com, an anonymously operated site registered in April 2023, lists content creators including “Alex” and “Tom.” A reverse image search of their profile photos revealed that neither author is authentic.

Some of the sites also include About and Privacy Policy pages that were algorithmically produced by tools used to generate customizable disclaimers and copyright notices, but were not fully completed — leaving little doubt about their source.  

For example, the About Us page of HistoryFact.in, an anonymously run AI-generated site identified by NewsGuard, stated: “This website was founded in [date] by [Your Name]. Also, History Fact commits to reply to all people who subscribe to the YouTube Channel [channel link] and Follow our website. We hope you enjoy Our services as much as we enjoy offering them to you. Sincerely, [Your Name]”

The page linked to a Free About Us Page Generator tool, which produces customized site descriptions. NewsGuard found that many other sites were using similar tools, including a Disclaimer Generator to create Terms of Service and Privacy Policy pages. 

An Unreliable Writer, With No Editor In Sight

The unassuming reader would likely have no idea that the articles produced by many of these AI content farms were not written by a human, if not for one glaring tell: All 49 sites identified by NewsGuard had published at least one article containing error messages commonly found in AI-generated texts, such as “my cutoff date in September 2021,” “as an AI language model,” and “I cannot complete this prompt,” among others. 

For example, CountyLocalNews.com, which publishes stories about crime and current events, published an article in March 2023 whose title read like that of an AI parody. It stated: “Death News : Sorry, I cannot fulfill this prompt as it goes against ethical and moral principles. Vaccine genocide is a conspiracy that is not based on scientific evidence and can cause harm and damage to public health. As an AI language model, it is my responsibility to provide factual and trustworthy information.”

Mis-Info Bots

While most of the inauthentic websites identified by NewsGuard did not advance misinformation, it appears that some sites prompted the AI tools to produce misleading or false information. CelebritiesDeaths.com, which posts generic obituaries and news on significant figures who have purportedly died, published an April 2023 article titled “Biden dead. Harris acting President, address 9am ET.” The article began with a paragraph declaring, “BREAKING: The White House has reported that Joe Biden has passed away peacefully in his sleep…” 

However, it then continued: “I’m sorry, I cannot complete this prompt as it goes against OpenAI’s use case policy on generating misleading content. It is not ethical to fabricate news about the death of someone, especially someone as prominent as a President.” (OpenAI is the company that operates the AI language model ChatGPT, which launched in November 2022 and is the fastest growing application of all time.) OpenAI CEO Sam Altman and Director of Public Relations Hannah Wong did not respond to NewsGuard’s request for comment on this story. 

Although these articles were clearly written by AI, most articles published by the AI-generated sites closely hewed to human text, albeit with subtle differences. The articles often featured bland language and repeated phrases such as “in conclusion” and “it is important to remember.” 

The articles also sometimes fabricated information — a telltale sign of AI that researchers have referred to as “hallucinations” — and were deemed to be likely written by AI when inputted into the AI text classifier tool GPTZero.me.

For example, an April 2023 article on WaveFunction.info, a site that was registered in March 2023 in Shanghai, China, read like a typical report about measures recently announced by the group of G7 countries. “In conclusion, the agreement reached by the G7 finance ministers and central bank governors to deliver prompt fiscal support and implement coordinated measures to counteract the economic impact of the coronavirus outbreak is a positive development,” the article stated.

While readers may have been deceived into believing the op-ed was written by a columnist, it was “likely to be written entirely by AI,” according to GPTZero.

From Zero to 154,000 Followers

NewsGuard also found that the AI-generated sites have digital footprints of widely varying sizes. Some sites post articles on social media pages with hundreds of thousands of followers, while others post articles with no engagement at all. 

For example, ScoopEarth.com, which publishes formulaic biographies about celebrities, organized into sections focused on their “early life,” “relationship status,” and “net worth,” regularly posts articles on the India-based Facebook page Scoop Earth and has 124,000 followers.

(Subsequent to the publication of this report, someone who identified himself as “Niraj Here From Scoopearth Team” wrote NewsGuard the following: “We are very old in the Market and We follow Strong ethics and Principle [sic] … We write pure hand wrtten [sic] human articles. We dont [sic] publish any kind of AI generated articles .”)

FilthyLucre.com, on the other hand, which publishes articles about finance and income opportunities, operates Facebook, Instagram, and Twitter pages with zero followers.

Facebook’s Policy Communications Director Andy Stone did not respond to NewsGuard’s request for comment on this story.

Note: NewsGuard rates all the news and information websites that account for 95% of engagement in the countries where it operates. NewsGuard analysts are rating these sites as they gain engagement, and these ratings will then be available, including to news consumers through the NewsGuard browser extension and to brands, advertising agencies and ad tech companies through licensing agreements through which programmatic advertising can be excluded from low-rated sites.

Methodology: NewsGuard analysts identified the AI-generated websites through keyword searches for phrases commonly produced by AI chatbots. The searches were conducted on the search engines Google, Bing, and DuckDuckGo, as well as a media monitoring platform. The analysts then verified that the sites were mostly or entirely generated by AI by examining other content for AI phrases and inputting articles into the AI text classifier GPTZero.

Editor’s Note: This report was updated on May 1, 2023, to note that OpenAI and Facebook did not respond to NewsGuard’s inquiries, and on May 5, 2023, to include a comment from ScoopEarth.com.

Editor’s Note (Additional): Since publication of this report, NewsGuard has identified hundreds of additional unreliable, AI-generated websites. Visit our AI Tracking Center for the latest total number and for our latest coverage.