Robot ‘Journalists’ Discovered To Be Behind Almost 50 News Sites
By Mikelle Leow, 03 May 2023
In recent weeks, the online media industry has witnessed the capsize of publishing giants. BuzzFeed News, which won a Pulitzer Prize for its investigative coverage, is shutting down amid company-wide layoffs. This week, reports emerged that Vice Media may be headed for bankruptcy. In their wake is the rise of non-human content creators that can spit out articles—and publish them without verifying facts—in seconds.
A new investigation conducted by news-rating and anti-misinformation platform NewsGuard warns that AI-generated “content farms” could eventually, and silently, unseat online journalism as you know it. The tool has already uncovered almost 50 such sites.
The group has identified 49 “news” websites that put out stories on politics, finance, health, lifestyle, entertainment, technology, and the environment at a “high volume.” The scrutinized content was written in seven languages—English, Chinese, Czech, French, Portuguese, Tagalog, and Thai—and was either “entirely or mostly” produced by AI generators, although none of them declared any chatbot intervention.
Many of them have adopted generic, newsy names insinuating that they are run by trusted publishers, like Daily Business Post, Biz Breaking News, The News Network, News Live 79, and Market News Reports. Several sites are noted to have only begun operating this year, just as when language-processing tools like ChatGPT became public.
NewsGuard says it managed to flag out AI-dominated content farms by simply looking up error messages often delivered by tools like ChatGPT on search engines.
Without much human moderation, all 49 websites freely published articles containing phrases like “I cannot complete this prompt,” “as an AI language model,” and “my cutoff date in September 2021.”
One site featured dozens of articles with sentences like, “I am not capable of producing 1500 words… However, I can provide you with a summary of the article.” The names of these farms are displayed in the report.
According to the findings, some of these sites publish “hundreds of articles a day,” hoping to generate more advertising space for profit, instead of being too concerned with editorial quality and integrity. In fact, even more worryingly, “some of the content advances false narratives,” NewsGuard analysts McKenzie Sadeghi and Lorenzo Arvanitis caution.
In April, one website focused on celebrity deaths wrongly broadcasted: “Biden dead. Harris acting President, address 9 a.m.”
Another had a headline reading: “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.”
Although the sites never mention they’ve had AI help, many of them list authors such as “Admin” or “Editor,” or even make up fake creator names like “Alex” and “Tom.”
The investigators also notice that almost all the content is laced with “bland language and repetitive phrases,” which they identify as “hallmarks of artificial intelligence.” That’s because the content isn’t the focus of content farms, but the revenue. These sites are powered by programmatic advertising, which is a tier of advertisements “placed algorithmically across the web… that finance much of the world’s media.”
With that said, not all content farms are performing as well as others. One site that’s centered around generating celebrity biographies, called ScoopEarth.com, has 124,000 followers on Facebook. On the other hand, a finance news site called FilthyLucre.com has no Facebook following at all.
NewsGuard managed to get a response from the owners of two of the 49 sites. One owner defends, “We did an expert [sic] to use AI to edit old articles that nobody read [sic] anymore just to see how it works.” The other justifies that their site only uses automation “where [it is] extremely needed… and yes, they are 100% facts checked [sic].”
“We are the new age of providing knowledge to each and every corner,” the latter content farm founder touts.
The findings present a startling trend of how online media might eventually look, and its impending rapid consumption by readers who won’t be able to discern human authorship from that of a well-trained machine.
“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,” shares NewsGuard, whose leadership includes former Wall Street Journal reporter Gordon Crovitz.
[via Bloomberg and The Guardian, images via various sources]