Content
summary Summary

NewsGuard has identified 713 AI-generated news sites that operate with minimal human oversight and frequently spread false claims.

Ad

These sites, which NewsGuard calls "Unreliable AI-Generated News and Information Websites" (UAINS), operate in 15 languages and typically pose as established news platforms. The number of UAINS detected has more than tripled from 217 in June 2023.

NewsGuard identifies these sites based on error messages or phrases from AI models such as "As an AI language model". Because the detection method is imprecise, expect a high number of undetected sites.

Monetizing misinformation

UAINs often have generic names such as iBusiness Day, Ireland Top News, and Daily Time Update, which appear to consumers to be established news sites.

Ad
Ad

However, according to NewsGuard, they actually publish "dozens and in some cases hundreds of generic articles" that are largely or entirely written by bots, rather than publishing traditionally written and edited journalism under human supervision.

The articles cover a wide range of topics, including politics, technology, entertainment, and travel. Sometimes they contain false claims, for example about political leaders.

Other false stories include celebrity deaths, fabricated events, and articles that recount old events as if they had just happened.

NewsGuard has also identified a Chinese government website that uses AI-generated text to spread the news that the U.S. is running a bioweapons lab in Kazakhstan that is infecting camels to endanger people in China.

According to NewsGuard, UAINs are a growing problem in the media landscape. They could lead to an increased spread of misinformation and disinformation that influences public opinion and political decisions. The sites make money from banner ads.

Recommendation

Interested parties can contact NewsGuard directly for the full list of domains and further details.

Other examples of harmful AI content on the web

Another example of AI fake news is the story orchestrated by the World of Warcraft community on Reddit. They invented the character "Glorbo" and presented it as a new addition to the game, prompting the AI-driven gaming news site "The Portal" to generate an extensive article about the fictional character.

Another site, "gameishard.gg," was able to reach an audience of millions overnights through Google channels such as Search and Discover with generic, partially false, or stolen AI-generated content about games.

There are also cases of individuals using AI-driven techniques to generate web traffic. SEO expert Jake Ward, for example, has successfully taken the principle of AI spam to the extreme.

Ad
Ad
Join our community
Join the DECODER community on Discord, Reddit or Twitter - we can't wait to meet you.

His campaign, which he calls an "SEO heist," brought massive traffic to his client's website, a software company called Causal, in a very short time with minimal effort. Ward used a GPT-4-based SEO tool to generate articles based on suggestions and submit them directly to the content management system.

Using this method, Ward published 1,800 articles in a matter of hours, resulting in 3.6 million visits and 13,000 keywords on the first page of Google since the project began.

Google became aware of the project when Ward published his results. Soon after, Causal's traffic returned to worse levels than before the "heist".

Support our independent, free-access reporting. Any contribution helps and secures our future. Support now:
Bank transfer
Summary
  • NewsGuard has identified 713 AI-generated news sites that frequently spread false information and operate with minimal human oversight. This compares to 217 in June 2023.
  • These unreliable AI-generated news sites (UAINS) are available in 15 languages and often masquerade as legitimate news platforms, but primarily publish articles written by bots.
  • According to NewsGuard, UAINS are a growing problem because they can contribute to the spread of misinformation and disinformation that influence public opinion and political decisions.
Sources
Online journalist Matthias is the co-founder and publisher of THE DECODER. He believes that artificial intelligence will fundamentally change the relationship between humans and computers.
Join our community
Join the DECODER community on Discord, Reddit or Twitter - we can't wait to meet you.