AI used for sophisticated advertising fraud

Author Moonshot.news

Published on: 03/10/2023

22% ($84 billion) of all online ad spend is lost due to ad fraud in 2023 and it is projected to reach over $170 billion in 5 years, a report from US analyst firm Juniper Research’s shows. Fraudsters capitalize on the use of AI by using algorithms to create bots and malware that can mimic human behaviour, the report says. 

Key findings:

• 22% of all online ad spend is lost due to ad fraud in 2023
• 30% of mobile ad spend is lost to ad fraud in 2023
• $172 Billion of ad spend is projected to be lost due to ad fraud by 2028
• $23 Billion per year can be recovered by using fraud mitigation platforms

“Data provided by popular ad platforms, such as Facebook and Google, provide an incomplete picture of the success of advertising campaigns,” says the company’s Senior Research Analyst, Elisha Sudlow-Poole. 

“That is, these platforms give an optimistic view of campaign efficiency, failing to distinguish between how many clicks or views originated from legitimate users compared to click farms or fraudulent bots.”

Juniper Research anticipates that, of the three online advertising types examined, search advertising will generate the largest fraudulent activity over the next five years. Online fraud is most likely to occur within browsers that have a smaller market share, the report says.

The focus on artificial intelligence shows also in the report on ad fraud:

“Fraudsters can capitalize on the use of AI by using algorithms such as ChatGPT to program advanced algorithms to create bots and malware that can mimic human behaviour.” 

“Mimicking this behaviour allows fraudsters to steal ad spend from advertisers by creating ad impressions and clicks that real users will not see. SIVT (Sophisticated Invalid Traffic) is more difficult than GIVT (General Invalid Traffic) to detect because fraudsters are actively changing patterns of attack to avoid detection.” 

The report says that to do this, these fraudsters are investing in AI to not only mask their illegal behaviours, but also detect opportunities where they can spoof valid traffic. 

“To combat SIVT, fraud mitigation providers are also having to invest in AI analytics that provide multi-point corroboration to detect, identify and analyse this traffic. However, SIVT (Sophisticated Invalid Traffic) uses AI to actively avoid detection from AI-based fraud mitigating frameworks; creating a ‘cat-and-mouse’ game involving a form of adversarial AI and resulting in a lack of return on investment for advertisers.”