The rise of AI-generated news websites is proliferating online, with nearly 50 news websites identified by NewsGuard as being almost entirely written by artificial intelligence software.
The rise of AI-powered content farms: Concerns of false information and clickbait articles
This new generation of content farms relies on low-quality websites that churn out a vast amount of clickbait articles to optimise advertising revenue. The concerns are that false information could serve as the basis for future AI content, creating a vicious cycle of fake news.
NewsGuard identified 49 websites in April 2023 that appear to be entirely or mostly generated by artificial intelligence language models designed to mimic human communication, spanning seven languages, including Chinese, Czech, English, French, Portuguese, Tagalog, and Thai. These sites often fail to disclose ownership or control and produce a high volume of content on various topics such as politics, health, entertainment, finance, and technology.
The websites publish hundreds of articles a day, with many of the sites saturated with advertisements indicating that they were designed to generate revenue from programmatic ads, just as the internet’s first generation of content farms operated by humans.
As AI tools become more powerful and widely available, the concerns that they could be used to conjure up entire news organisations have now become a reality. NewsGuard sent emails to the 29 sites in the analysis that listed contact information, and only 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.
The owner of Famadillo.com, a site that has published numerous AI-generated product reviews attributed to “admin,” denied that the site used AI in a widespread manner, despite exchanging hard-to-comprehend emails with NewsGuard.
Meanwhile, 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, confirmed that they use automation where necessary and assured that their content is “100% facts checked” to prevent the creation of false information. Ingale did not elaborate on the site’s use of AI, but claimed that the site’s content is published manually under human supervision.
The issue at hand highlights the increasing challenge of distinguishing between human-generated content and that produced by automated bots. By the end of a full day of examining the content on such sites, it was even more difficult to discern what was authentic and what was not. Furthermore, the impact of these AI-generated websites varies significantly, with some sites reaching audiences of tens of thousands on Facebook while others struggle to gain any traction at all.
Critical thinking against AI news
Various measures can be taken to identify and prevent the spread of fake news. One such method is to employ critical thinking skills to evaluate the information presented in news articles and to scrutinise their sources.
Additionally, awareness of the potential biases and motivations behind news sources can be crucial in discerning whether the information is accurate or intentionally misleading. By exercising these safeguards, readers can help prevent the spread of false or misleading information and promote the dissemination of accurate, trustworthy news.
Generic Names run rampant in AI-generated news sites
NewsGuard has identified 49 AI-driven news sites with seemingly benign and generic names that suggest established publishers such as Biz Breaking News, News Live 79, Daily Business Post, and Market News Reports. However, upon closer inspection, these sites appear to primarily consist of content that has been rewritten or summarised from other sources using AI algorithms.
One example is the website BestBudgetUSA.com, which was anonymously registered in May 2022 and appears to summarise or rewrite articles from CNN. The site often features phrases commonly produced by generative AI in response to prompts such as “I am not capable of producing 1500 words…” followed by a summary of the article and a link to the original CNN report. The presence of these phrases suggests that these sites operate with little to no human oversight.
Moreover, many of the AI-generated articles are credited to anonymous authors such as “Admin” or “Editor,” while others feature fake author profiles. For example, the website HarmonyHustle.com lists content creators including “Alex” and “Tom,” but a reverse image search of their profile photos revealed that neither author is authentic.
Left-over AI-generator jargon
According to NewsGuard, the articles generated by many of the 49 AI-driven content farms are often indistinguishable from those written by humans. However, there is one telltale sign that exposes their AI origin. All 49 sites identified by NewsGuard have published at least one article containing error messages that are commonly found in AI-generated texts. These error messages include phrases such as “my cutoff date in September 2021,” “as an AI language model,” and “I cannot complete this prompt.”
For instance, CountyLocalNews.com, which focuses on publishing stories related to crime and current events, released an article in March 2023 with a title that appeared to be a parody of AI-written articles. The title read: “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.”
Prompts requesting misinformation
While most of these websites did not publish misinformation, some prompted the AI tools to produce misleading or false information. For instance, CelebritiesDeaths.com, a website that publishes obituaries and news on famous people who have died, published an article in April 2023 claiming that President Biden had passed away in his sleep. However, the article was abruptly cut short with a message stating that generating such misleading content was unethical and violated OpenAI’s use case policy.
Although these articles were clearly written by AI, they closely resembled human-written text, albeit with subtle differences. The articles often contained bland language and repeated phrases such as “in conclusion” and “it is important to remember”. However, they sometimes fabricated information, a sign that researchers have referred to as “hallucinations”. Such articles were identified as likely written by AI when analysed by the AI text classifier tool GPTZero.com.
For example, WaveFunction.info, a website registered in March 2023 in Shanghai, China, published an article in April 2023 about a recent agreement reached by the G7 finance ministers and central bank governors. While the article read like a typical report, it contained subtle differences that suggested it was generated by AI.
Although these articles are difficult to detect, they pose a significant threat to the credibility of news content. NewsGuard’s findings indicate the need for greater transparency in the production and dissemination of news content, especially in the era of AI-generated text.
Join the AIBC discussion in São Paulo, Brazil
AIBC Americas in São Paulo, Brazil is a must-attend conference for anyone passionate about the future of technology in the Americas. The event offers a unique opportunity to network with industry leaders and participate in KOL-led conferences and innovative expos. Stay informed about the latest news and updates by visiting our website and secure your spot at this exciting event by registering early.