They're Giving It All - Tech That Detects AI-Generated Clickbait

Tools and chatbots are being used to generate massive amounts of low-quality content solely for the purpose of attracting programmatic ad spend.

Targeting signals are being deprecated and cookies are crumbling. Everything is connected. It creates tension between content, sentiment, and more.

Peer39 developed a tool to exclude these clickbait sites, disabling the ability for ads to run. The focus sits in the broader advertising theme — Made for Advertising — that looks at relevant, suitable, and other ways to bid on advertising. The company started to look at the content. There was a lot of chatter in the market about companies acting like content farms.

“You might not see the company name, but there were thousands of new pages being produced without writer names,” said Peer39 CEO Mario Diez. “Then we looked at trends that marketers were concerned about like what will happen to the web when artificial intelligence generates a ton of content — how will they understand what’s good and bad.”



Peer39 wanted to identify the bad actors or publishers making mistakes, errors, in content generated by large language models (LLMs).

Since the media buying-community depends on Peer39 to provide advertising intelligence in an environment, Diez decided the company should find the really bad actors who are beginning to use AI to create harm.

Similar to brand safety or contextual categories, Peer39 launched an AI Generated Content Category, giving programmatic advertisers the power to exclude these clickbait sites from their campaigns. It’s part of the broader Peer39 platform that falls into the quality suite of controls.

He said the limited editorial oversight is creating a trap for advertisers, with pages that often exploit metrics like viewability and clicks in order to capture programmatic demand, while providing little value to the advertisers.

Peer39 monitors about 2 billion URLs daily. Detection models were built that scan the web for oddities and accuracy.

Peer39 research into AI-generated bot farms finds that viewability scores are incredibly high for these sites, often in the range of 80% to 90%.

As a result, programmatic buying platforms will often optimize toward this content. Viewability is high, but attention is low. These ads game the market through high viewability and low attention metrics and bad consumer experiences.

The problem is that Peer39 rated these pages as 40% to 60%, well below the average benchmarks, proving that evaluating impressions with multiple metrics is required for digital marketers. The data was recently released in its Attention Index.

Viewability is a measure of ads in the view of users. The additional inputs for Attention provide a deeper understanding if a user is interacting with ads and the overall environment.

The Attention Index is a metric available in Analytics that can guide advertisers toward improving performance using KPIs. With the daily updates of categories, domain-level adjustments automatically happen. For specific Contextual Categories, Analytics shows the Attention Index, which can aid buyers in optimization.

The Index is based on the scores of all placements on a domain compared to the average of all domains, according to Peer39.

These scores are predictive, based on past placements, but in some cases, an individual category may fall below or be well above the minimum of the targeted Peer39 Attention Index tier.

Peer39 research also found that the most popular content topics for AI Generated Bot Farms are relationships, celebrities, food and beverage, healthy eating and style and fashion. 

Next story loading loading..