It has been a hot ad fraud summer at DV. In July alone, the DV Fraud Lab detected 34% more bot scheme variants than we did last July. We flagged nearly two billion requests from the sophisticated ViperBot scheme, which spoofs premium video inventory on mobile and CTV devices. And we identified a significant spike in an adware scheme targeting Android devices.
The DV Fraud Lab attributes these dramatic surges in activity to agentic AI, which is accelerating the pace at which fraud schemes scale, spike and multiply. But DV’s fraud fighters are moving just as quickly to protect our advertiser customers’ campaigns. Our secret weapon? Also agentic AI.
DV is using our massive proprietary data and fraud expertise to build AI skills that investigate along with us, like research buddies that never get tired.
Built by DV’s Fraud Analytics Engineering Group in collaboration with the Fraud Research Engineering Team, the agents learn from human domain experts to dynamically collect and analyze evidence and prioritize leads at superhuman speed. In a matter of hours, they can proceed from fraud detection to evidence-gathering to mitigation, a process that might take a human a couple of days.
These AI game-changers help the Fraud Lab mitigate dozens of schemes, which together account for hundreds of millions of impressions, every day.
DV’s AI Agents in Action
To demonstrate, here’s how a DV mobile fraud agent helped us quickly respond to a fraudulent mobile app scheme, “Puppet Smash.”
To determine whether the app was fraudulent, a human investigator would first need to analyze its underlying code and configuration to spot red flags that signal malicious intent. The work is painstaking and requires reasoning and judgment, not just pattern-matching.
The Fraud Lab’s AI agent does much of this analysis for us. It automatically detects evasion methods in the app’s code and circumvents them to reveal the app’s true nature. Then it tests the app and documents its behavior with a screenshot and a report.
AI agents also helped DV uncover the adware scheme we’ve named AfterCall. In an AfterCall attack, a malicious app dodges built-in platform safeguards, tricks a smartphone user into granting overlay permissions, then begins firing ads every time the user ends a phone call. Our agents have helped us swiftly respond to this proliferating threat.
DV’s Agentic Fraud Lab Has You Covered
As of early 2026, AI agents, tools, skills and workflows are fully integrated into DV Fraud Lab operations as part of DV Neura™, the cognitive engine that powers our platform.
DV has shut down thousands of web, in-app and streaming/CTV fraud properties and schemes over the past few years. By providing enhanced fraud detection coverage that keeps pace with the growth in fraud attempts, our AI agents help keep the industry, and your ad dollars, safe.
Next Steps
Contact a DV representative.
Related Topics
Recent Posts
