ADEX report: AI involved in 52% of detected ad fraud schemes in 2026; AI shopping agents flagged as next target
ADEX, an AI-powered traffic validation and anti-fraud platform within AdTech Holding, reports that AI played a significant role in 52% of advertising fraud schemes it identified between January and September 2026. The data comes from aggregated moderation of eight advertising platforms, covering 1.2 million campaigns and 4.9 million creatives. Among AI-involved schemes, bots impersonating humans appeared in 64%, cloned brands and fake endorsements in 53%, and deepfakes in 11% (categories overlap). ADEX forecasts that in 2027 fraudsters may increasingly target AI shopping agents, for example through prompt injection and cloaking. This is a vendor forecast, not an attack category currently measured in its data.
Detalhes técnicos
- AI lowers the cost of producing and scaling fraudulent content, including convincing imitations of news outlets and financial institutions, and chatbots that handle victim conversations around the clock.
- Brand cloning and fake endorsements appear in investment and blockchain scams, fraudulent e-commerce stores, and fake security alerts that imitate technology brands.
- Revised versions of rejected fraudulent campaigns reappeared within hours, sometimes on the same day.
- Forecast threat: prompt injection against AI agents via invisible page elements, metadata, product images, or text addressed to AI assistants. Attackers could also serve different content to humans and AI agents, adapting existing cloaking techniques.
Remediação
- Platforms should verify that a visitor can prove who it is, that its behavior matches its stated task, and that its sessions end in real outcomes such as purchases or bookings.
- Platforms should watch for sites that show AI agents one page and humans another.
- Advertisers should monitor for their brand in others' ads, check where clicks actually land, and ask technology partners how they distinguish good agents from bad automation.
Linha do tempo
-
ADEX report: AI involved in 52% of detected ad fraud schemes in 2026; AI shopping agents flagged as next target
ler a análise desta fonte ocultar
ADEX, an AI-powered traffic validation and anti-fraud platform within AdTech Holding, reports that AI played a significant role in 52% of advertising fraud schemes it identified between January and September 2026. The data comes from aggregated moderation of eight advertising platforms, covering 1.2 million campaigns and 4.9 million creatives. Among AI-involved schemes, bots impersonating humans appeared in 64%, cloned brands and fake endorsements in 53%, and deepfakes in 11% (categories overlap). ADEX forecasts that in 2027 fraudsters may increasingly target AI shopping agents, for example through prompt injection and cloaking. This is a vendor forecast, not an attack category currently measured in its data.
Technical Details
- AI lowers the cost of producing and scaling fraudulent content, including convincing imitations of news outlets and financial institutions, and chatbots that handle victim conversations around the clock.
- Brand cloning and fake endorsements appear in investment and blockchain scams, fraudulent e-commerce stores, and fake security alerts that imitate technology brands.
- Revised versions of rejected fraudulent campaigns reappeared within hours, sometimes on the same day.
- Forecast threat: prompt injection against AI agents via invisible page elements, metadata, product images, or text addressed to AI assistants. Attackers could also serve different content to humans and AI agents, adapting existing cloaking techniques.
Remediation
- Platforms should verify that a visitor can prove who it is, that its behavior matches its stated task, and that its sessions end in real outcomes such as purchases or bookings.
- Platforms should watch for sites that show AI agents one page and humans another.
- Advertisers should monitor for their brand in others' ads, check where clicks actually land, and ask technology partners how they distinguish good agents from bad automation.