On AlaskaCommons.com, writer “Chelsea Waller” says she is a “dedicated journalist with a passion for exposing the truth.” One of hers stories reads: “As an AI language model, I can’t predict whether or not John Wick 4 will be a finale to a saga that will be remembered for years to come.”
The website is just one of NewsGuard-flagged news sites filled with spammy, low-quality content generated using artificial intelligence tools, where ads for major brands now appear, according to a report by NewsGuard. The study was the first indicated by MIT Technology Review and includes brands such as Citigroup and Subaru.
More than 140 brands placed ads on sites NewsGuard describes as untrustworthy, AI-generated news sites — probably without even knowing the ads were there. NewsGuard discovered that hundreds of ads appeared on the websites through programmatic Google ads, a system that automatically places ads on sites rather than the advertiser choosing the placement. In turn, the AI-generated websites benefit by displaying ads next to cheap content.
Some of the sites monitored by NewsGuard publish thousands of stories per week, such as World-Today-News.com, including Article copied and pasted by The New York Times and stories along headlines read: “Sorry, I am an AI language model and I cannot rewrite the given title because it is not clear and is not a recognizable language. Please provide a clear and readable title.” Ads appeared on both stories at the time of writing.
On one website, MedicalOutline.com, advertisers included Subaru, Citigroup and GNC, according to MIT Technology Review. The site provides incorrect health information such as “What are 5 Natural Remedies for ADHD?” and “Can Lemon Cure Skin Allergy?”
Google has regulations around where ads can be placed – the company prohibits ads on spammy auto-generated content, as well as material copied from elsewhere, with the same policy penalizing spam in search results or lowering rankings. Google did not immediately respond to a request for comment.