Slop is sloppy, it seems.
That’s according to a new report from DoubleVerify, which has uncovered a fraud scheme dubbed AutoBait — a network of more than 200 AI-generated websites. The network, according to DV, “has generated tens of millions of impressions, which unprotected advertisers are unknowingly paying for.”
One of the more fascinating aspects of the scheme is that, in the rush to build the network, the fraudsters left what amounts to a prompt graveyard in the site JavaScript. The embedded prompts reveal everything from the model used to generate the network to the types of imagery and headlines the pages are designed to feature.
Why This Matters:
The code with the prompt – which can be seen here – gives AI slop researchers some interesting new depth on ohw these bad actors operate. Here’s more from DV:
AutoBait’s prompts are meticulously crafted to generate slide show-style clickbait that misleads users and manipulates engagement. The prompts instruct the LLM to frontload the first few slides with “the most sensational or shocking points — anything that stops someone mid-scroll.”
Headlines must be “ultra-literal, e.g., ‘A mole with a funny shape,’ ‘A spot that itches nonstop,’ ‘A sore that won’t heal.’
The body text has to “inject real emotion (fear, anger, shock, relief) into every paragraph.”
The prompt even includes specific manipulation tactics:“Ask a direct question, hint at danger or relief, and convey urgency.”
“Be highly specific: name colors, objects, or settings exactly. Paint a scene someone could immediately picture and photograph.”
The company adds that, at a basic level, the entire system is designed to exploit human psychology and behavior. The prompts generate AI content meant to trigger emotional responses, using fear and sensationalism to drive engagement. Clickbait, basically.
It’s also worth noting the scale of the network. According to DV, saying that, in a month, a network like AutoBait can pump out “tens of thousands of pages with millions of ad-serving opportunities.” “If even a fraction of those impressions are filled,” says DV, “the scheme generates significant revenue.”
Experts React:
A couple of weeks ago, the CTO of DV, Nisim Tal, wrote this in a piece for The Media Leader, which feels especially relevant now:
AI will soon represent a meaningful share of what people watch, read, and engage with. I also believe and predict it will become the dominant form of content online.
We’re already seeing this shift happen. AI-generated articles and human-written articles on the open web are now roughly equal in volume. That equilibrium won’t last. As generative tools improve and adoption accelerates, AI text will outpace human text within a year, with visuals likely following soon after.
Our Take:
AI slop is a huge — and growing — problem. Per DV’s CTO, AI-generated content is set to eventually overtake human-created content online. That means this isn’t a problem that inclusion lists alone, for example, can solve.
Addressing it will require safeguards across the ecosystem: verification, stronger protections built into platforms (both on the open web and in walled gardens), more supply-side curation, and better AI literacy among agencies and media buyers.
AdTechRadar is owned by Chris Harihar, who leads PR at Mod Op. DoubleVerify is a Mod Op client.
