It is clear that low-quality news sites using chat AI for profit-making purposes are rampant
Rise of the Newsbots: AI-Generated News Websites Proliferating Online - NewsGuard
https://www.newsguardtech.com/special-reports/newsbots-ai-generated-news-websites-proliferating/
AI Chatbots Have Been Used to Create Dozens of News Content Farms - Bloomberg
Group identifies dozens of news sites created by AI chatbots: report | The Hill
AI Content Farms Use OpenAI's ChatGPT for Fake News Stories
https://gizmodo.com/chatgpt-ai-fake-news-stories-content-farms-newsguard-1850391104
In April 2023, NewsGuard identified 49 websites using chat AI to create low-quality content. These websites operate in seven languages: Chinese, Czech, English, French, Tagalog, and Thai. It is believed that some or all of the content on these websites is generated by chat AI. These websites generate large amounts of content related to a wide variety of topics, including politics, health, entertainment, finance, and technology, with some sites publishing hundreds of articles per day.
According to an investigation by Bloomberg, these websites sometimes appear as generic news sites, such as 'News Live 79' and 'Daily Business Post.' While not all of the website content is generated by chat AI, none of the sites explicitly claim to use ChatGPT or Google Bard.
NewsGuard has published some of the articles published on these websites, and has noted that many of the articles are summaries or rewrites of news published by well-known news outlets such as CNN.
One low-quality chat AI site, CelebritiesDeaths.com, has also been found to contain false information about President Joe Biden . The site featured an article titled 'President Biden Has Died,' which claimed that Vice President Kamala Harris had taken over as president. However, the article still contained an error message that read, 'Sorry, we cannot complete this prompt because it violates OpenAI's policy regarding the generation of misleading content. It is unethical to fabricate news about the death of someone, especially a prominent figure like the president.'
Furthermore, an article posted by the website 'BestBudgetUSA.com' included a message verbatim output from a chat AI that read, 'I cannot generate sentences longer than 1500 words. However, I can provide a summary of the article.'
According to NewsGuard's research, most of the articles on these news sites contain 'bland language' and 'repetitive phrases,' which are characteristic of AI-generated text. Furthermore, these sites are flooded with advertisements, making them likely to be typical clickbait articles.
'Using AI models known to fabricate facts to create sites that look like news sites is a fraud against journalism,' said NewsGuard CEO Gordon Crovitz. 'AI developers like OpenAI and Google should train their models to be incapable of fabricating news stories.'
'The recent months have seen a number of powerful AI announcements and publicity, making real concerns that media researchers once speculated about: that AI could be used to misrepresent news,' Crovitz said.
NewsGuard contacted 29 of these websites, and two acknowledged using AI. Of the remaining 27, two did not adequately respond to NewsGuard's inquiries, 17 did not respond, and eight used invalid email addresses.
In response to Bloomberg's question about whether AI-generated websites violate our advertising policies, a Google spokesperson said, 'When enforcing our advertising policies, we focus on the quality of the site, not how the content is created, and if we find violations, we will suspend or remove ads from the site.'
NewsGuard warns that 'the 'discreet reader' would never know that the articles published on these websites were written by AI.'
Related Posts:
in AI, Web Service, Posted by log1r_ut