DoubleVerify gen AI fuels surge in new Ad fraud schemes and variants

The DV Fraud Lab observed a surge of over 20% in new fraud schemes, with a nearly 60% rise in audio and video streaming fraud schemes and variants. DoubleVerify (“DV”), the leading software platform for digital media measurement, data, and analytics, previewed its eighth annual Global Insights Report. This report examines media performance and trends across more than one trillion impressions from over 2,000 brands in 100 markets. The findings highlight a significant year-over-year increase in new fraud schemes, as fraudsters increasingly target platforms like streaming audio and Connected TV (CTV).

The DV Fraud Lab reported a 23% increase in the number of novel fraud schemes and variants in 2023 compared to the previous year. Furthermore, advertisers who were not protected experienced a fraud/SIVT (Sophisticated Invalid Traffic) violation rate that exceeded 15%. These violations encompass bot fraud, site fraud, app fraud, compromised devices, nonhuman data centre traffic, and injected ad events, all of which illegitimately inflate online advertising impressions or data events to generate revenue.

Generative artificial intelligence is a significant contributor to the rise. One of the discoveries made by DV is that Gen AI makes it simpler to fabricate data patterns.

Roy Rosenfeld, Head of the DV Fraud Lab mentioned that bot fraud, for example, frequently targets streaming settings such as CTV and in these contexts, fraudsters rely on false impression data to make it appear as though advertisements are being delivered to real people. With the use of artificial intelligence, attackers are able to build thousands of user agents that appear to be authentic. These user agents retrieve and facilitate end-user interaction with web content. Subsequently, attackers could employ artificial intelligence to mimic human behaviour, hence complicating the detection of patterns in bot traffic activity.

Not only are bad actors employing artificial intelligence to expand their schemes, but DV also discovered that plans to create false impressions are developing at a faster rate. Twenty-six percent more versions of the typical bot fraud scheme that targets streaming platforms were created in 2023, bringing the total number of streaming fraud schemes and variants to 58 percent higher than the previous year.

Artificial intelligence also makes it simpler for dishonest individuals to construct new websites for shell firms and publish new applications that contain dozens of bogus evaluations that give the impression that they have an audience. These kinds of deceptive strategies were found to be used extensively in mobile app schemes, according to investigations into those schemes. This kind of deception is becoming easier to produce as a result of artificial intelligence. 

According to Rosenfeld, “fortunately, excellent artificial intelligence may be utilised to attack poor AI.” DV, for instance, makes use of AI in order to handle enormous volumes of data on a regular basis and evaluate the quality of impressions. The processing of data in real time is made more efficient by machine learning. After then, artificial intelligence is able to recognise patterns and trends that would be unnoticeable to a human analyst who is not made use of modern AI approaches. Because of the increasing sophistication of fraudsters through the use of AI, it is absolutely necessary for us to also use AI in order to defend ourselves from their scams.