"The operation, which DoubleVerify called "AutoBait," created a range of unsophisticated websites featuring slideshows with attention-grabbing photos and headlines designed primarily to capture digital advertising revenue. The articles often appeared nonsensical, reinforcing that the goal was not reader loyalty, but rapid ad impressions."
"For instance, the model was told to front-load the first few slides with "the most sensational or shocking points - anything that stops someone mid-scroll." Headlines needed to be "ultra-literal" and the body text needed to "inject real emotion (fear, anger, shock, relief) into every paragraph.""
"Many of the domains previously hosted legitimate content, but the owners let their registration lapse. This allows scammers to capitalize on existing domain history and traffic, Gilit Saporta, senior director of fraud and quality analysis at DoubleVerify, told Axios."
A coordinated fraud operation called AutoBait created more than 200 templated websites using large language model prompts to generate sensationalized content designed primarily for capturing digital advertising revenue. These fake websites featured slideshows with attention-grabbing photos and nonsensical articles, indicating the goal was rapid ad impressions rather than genuine readership. Fraudsters provided specific LLM instructions to front-load shocking content, use ultra-literal headlines, inject emotional language, and generate realistic-looking photos. Many domains were previously legitimate sites whose registrations lapsed, allowing scammers to exploit existing domain history and traffic. This fraud method represents an evolution of pre-AI scam tactics, now enhanced through automated content generation.
Read at Axios
Unable to calculate read time
Collection
[
|
...
]