A recent court testimony has revealed that Google can continue training its search-specific AI models using web content, even when publishers have opted out of AI training.
During a trial examining Google’s search dominance, Eli Collins, Vice President at Google DeepMind, confirmed that while publishers can restrict their content from being used by DeepMind’s AI models, Google’s search division can still utilize the data for its AI-powered search features.
The revelation has sparked concerns among website publishers, who argue that Google’s AI-generated summaries at the top of search results discourage users from visiting independent websites, potentially impacting their revenue.
The U.S. Department of Justice (DOJ) is pushing for measures to restore competition in online search, including restrictions on Google’s AI practices and a proposal for the company to sell its Chrome browser and share key search data.
A document presented in court, titled “Search GenAI Gemini v3”, revealed that Google removed 80 billion out of 160 billion tokens—snippets of content—after filtering out material that publishers had opted out of allowing Google to use for AI training.
Despite the controversy, Google maintains that publishers can manage their content visibility in search results using the robots.txt web standard. However, critics argue that opting out of AI training should not require publishers to completely remove their content from Google Search.
The case continues to unfold, with regulators weighing potential antitrust actions against Google’s search and AI dominance.
Stay tuned for further updates on this developing story.
