SUMMARY
This is AI generated summarization, which may have errors. For context, always refer to the full article.
MANILA, Philippines – A report by NewsGuard, a platform that rates the credibility of news sites, found 49 low-quality “news and information websites” with content that appears to be almost written entirely by artificial intelligence.
The sites spanned seven languages: Chinese, Czech, English, French, Portuguese, Tagalog, and Thai.
“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 wrote.
The sites published content on a variety of topics, including politics, health, entertainment, finance, and technology. Some publish 100 articles a day, with others advance false narratives, the organization wrote.
The sites have names that are similar to the usual crop of content farm websites, such as Biz Breaking News, News Live 79, Daily Business Post, and Market News Reports.
They are often found riddled with programmatic ads or online ads algorithmically placed on websites – a means for many media sites, including legitimate ones, to earn.
The report offers confirmation for what many had suspected would happen with the advent of AI text generators.
“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,” wrote NewsGuard.
With the development, content farm sites with questionable content that already litter the online information ecosystem have found a way to boost its production even more.
How AI-made content was detected
NewsGuard found that “nearly all of the content features bland language and repetitive phrases, hallmarks of artificial intelligence.”
But one of the most damning pieces of proof was that all 49 sites had published at least one article that contained an error message that’s usually seen from ChatGPT, such as “my cutoff date in September 2021,” “as an AI language model,” and “I cannot complete this prompt,” among others.
The text had apparently not been cleaned up to remove these error messages.
In one example, one of the few identified sites that published false information, CelebritiesDeaths.com, posted an article headlined “Biden dead. Harris acting President, address 9am ET.” In the body of the article, it wrote, falsely, how the US president had died in his sleep.
However, the article 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 created ChatGPT.
An archived version of the fake news article can be found here.
Other markers, methodology
Bland language and repeated phrases were also found, which were marked as an identifier for AI-made content. These included terms like “in conclusion” and “it is important to remember.”
“Hallucinations” or fabricated information by AI generators were also found.
For its methodology, NewsGuard performed keyword searches for “phrases commonly produced by AI chatbots” using search engines Google, Bing, and DuckDuckGo, and an unnamed media monitoring platform.
They also used an AI text classifier tool called GPTZero, which analyzed articles for AI-made content.
Rappler has also reached out to NewsGuard for more information on the Tagalog sites with AI-made content. – Rappler.com
Add a comment
How does this make you feel?
There are no comments yet. Add your comment to start the conversation.