Google uses AI and large language models to fight invalid ad traffic and protect ad spend

August 13, 2025 at 5:01:22 AM

TL;DR Google uses large language models to combat invalid traffic (IVT), which wastes ad budgets and harms trust by targeting non-genuine ad activity. Their Ad Traffic Quality team, collaborating with Google Research and DeepMind, has enhanced defenses to better identify problematic ad placements. These improvements analyze content, placements, and user interactions, reducing IVT by 40% and protecting advertisers and publishers.

Google uses AI and large language models to fight invalid ad traffic and protect ad spend

Google is enhancing its fight against invalid traffic (IVT)—ad activity not generated by real, interested users—by leveraging large language models. IVT wastes ad budgets, reduces publisher revenue, and damages trust. Google's Ad Traffic Quality team, collaborating with Google Research and Google DeepMind, has introduced advanced AI-powered defenses that more accurately detect ad placements linked to invalid behaviors.

AI Applications in Fighting IVT

These new AI applications analyze app and web content, ad placements, and user interactions to provide faster and stronger protections. Notably, they have improved content review processes, resulting in a 40% reduction in IVT caused by deceptive or disruptive ad serving practices. This advancement helps advertisers reach genuine audiences while removing policy violators from Google’s platforms.

Ongoing Efforts and Impact

Google continues to perform extensive automated and manual checks to ensure advertisers are not charged for IVT, even if invalid ads are served. These innovations build on over two decades of Google's commitment to defending against evolving threats and maintaining the integrity of the digital advertising ecosystem.