Posted
April 22, 2024
In our years of protecting publishers from malvertising we know first hand that not all ads are what they appear. In this blog post, we’ll delve into the world of deceptive non-compliant ads discussing how our Boltive AI Engine is helping combat this problem.
The most basic definition of a compliant ad is one that adheres to a site’s accepted policy, while non-compliant ads are ones that break those rules. These violations can occur due to the ad content itself, such as alcohol ads on sites that prohibit them, or ads with negative sentiment, such as antagonistic political campaign ads. Some of the most concerning are ads that impersonate or “hijack” a well-known brand, a phenomenon called Brandjacking. This threat revolves around bad actors being able to exploit a brand’s goodwill and good reputation to engage in all kinds of scams.
Our Boltive AI Engine, developed in Fall 2023, now battles all types of ad content issues by utilizing sophisticated techniques to track down these non-compliant ads. By analyzing vast amounts of data, our AI models spot patterns and anomalies that human moderation might miss. See our earlier post “Ad Security Reimagined: Exploring Boltives Artificial Intelligence Techniques” for a deep dive into how our technology works.
When it comes to Brandjacking, the Boltive AI Engine serves as a great tool for identifying the misuse of a brand by bad actors. The system is remarkably good at identifying sentiment in ads, and by adding this new tool to their arsenal brands are able to take proactive measures to protect their reputation and maintain customer trust.
Brandjacking has become a recurring problem for the TV show “Shark Tank”. Bad actors are leveraging the popularity of the TV series to create ads for fictitious companies. These ads prey on the show’s core approach of finding great new products. For instance, USA Today reported a case in 2021 where a fake Shark Tank ad misled viewers while another instance involved AI-generated spoof ads that mimicked the show’s format.
Recently, our AI uncovered a fictitious Shark Tank ad claiming to stop tinnitus. Screenshots of the fake ad and its landing page reveal the extent of this misrepresentation.
Below is an example of Brandjacking that our team uncovered while reviewing ads that our AI model flagged in a query for misleading health claims for tinnitus. The malicious campaign consisted of a fraudulent ad that took viewers to an also fraudulent landing page.
As seen in Image 1, the use of the “Shark Tank” name is leveraged to draw readers to the ad based on the TV series’ brand reputation.
Once the ad is clicked, the viewer is redirected to a landing page that not only spoofs the Today show logo in its title (as seen in Image 2), but also continues to reference the Shark Tank fraudulent content.
Deceptive ads span every category we track, and with an election year underway we are already seeing heavy usage of misleading messaging in political ads. Examples include anti-Biden and anti-Trump ads from various Political Action Committees with references to major issues including the Israel-Hamas conflict. To help our customers better address the poor customer experience these ads drive, we are using our AI models to differentiate between ads with negative and positive messages. To do this, we are creating a new category that specifically blocks Negative Political ads while allowing standard campaign ads with positive messages to run.
With bad actors continuing to evolve and election season really ramping up, transparency and control are paramount. Our Boltive AI Engine stands ready to unmask deceptive ads and protect publishers. To witness it in action, book a demo today!
