AI Fuels Sophisticated Ad Fraud as Cloaking Declines, Report Finds
A new report reveals AI is now central to over half of ad fraud schemes, with tactics shifting from cloaking to AI-driven deepfakes, cloned brands, and sophisticated bot operations.
The ad fraud landscape has undergone a significant transformation in 2026, with artificial intelligence emerging as a dominant force in malicious schemes. The "State of Ad Fraud 2026: ADEX Trust & Safety Report" indicates that 52% of distinct fraud schemes now leverage AI-generated or AI-assisted assets, marking a substantial departure from previous years. This shift is detailed in an analysis of over 1.2 million campaigns and 4.9 million creatives across eight ad platforms.
While cloaking techniques, which previously accounted for approximately 75% of advertiser suspensions, have seen a notable decline to around 45%, new AI-powered threats have risen to prominence. The report highlights that 64% of AI-involved schemes employ bots designed to mimic human visitors, while 53% utilize cloned brands and fake endorsements to enhance their deceptive impact. Furthermore, 11% of these AI-driven schemes are now incorporating deepfakes, adding a new layer of sophistication to impersonation and manipulation.
Beyond AI's direct involvement in asset creation and bot operations, the report identifies other evolving threats. Malicious iOS scripts have emerged as a significant concern, accounting for roughly 20% of suspensions, indicating a growing focus on Apple's mobile ecosystem. Additionally, AI-powered messenger phishing, particularly on platforms like WhatsApp and Telegram, is becoming a potent tool. The report notes that advanced language models enable single operators to manage hundreds of convincing conversations simultaneously across multiple languages.
Infected domains have also become a primary rejection reason, rising to become the top cause from the second quarter onwards. Scareware tactics, such as "You won a smartphone" scams, have also emerged as a distinct category, comprising 6% of rejections in the third quarter. This indicates a diversification of fraudulent content designed to lure unsuspecting users.
The overall volume of rejected or blocked campaigns has decreased by about a third since the first quarter. However, this reduction does not signify a decline in ad fraud activity. Instead, operators are shifting their strategy from mass-produced, low-quality campaigns to fewer, more meticulously crafted operations. "One well-made scam with a cloned brand page does more damage than a hundred crude ones," stated Sergey Martianov, CPO at ADEX, underscoring the increased potency of these targeted attacks.
Geographically, Tier-1 markets (US, UK, Canada, Australia, Germany) consistently show a high risk of rejected campaigns. South Asia has emerged as a high-risk region from the second quarter onward, while Turkey was prominent in the first half of the year, and Latin America saw increased risk in Q1. East Asia also became a high-risk area in Q3.
Looking ahead, ADEX predicts several trends for 2027. These include the development of ads specifically designed for AI agents, the blurring of lines between bot and human traffic, end-to-end AI generation of fraud operations, increased reliance on compromised domains and messenger funnels for "borrowed trust," and the migration of scareware tactics to connected TV, in-app environments, and AI assistant interfaces. Prompt injection is also identified as a future threat vector.