Google’s AI search policy leaves publishers in a bind, court papers show

New Delhi: Google’s integration of artificial intelligence into its search engine is raising alarms among publishers, as court documents reveal the tech giant deliberately chose not to offer them a way to opt out of AI training while still appearing in search results.

According to news reports, the internal documents, revealed during an ongoing antitrust trial, show that Google considered giving publishers more control, such as asking for consent or letting them opt out of having their content used to train AI models. But the company decided against it, citing technical challenges and costs. According to a memo by Chetna Bindra, a senior Google Search executive, creating separate models for each search feature would have required heavy hardware investments and slowed down development.

Instead, Google reportedly made a quiet policy change, without public notice, that allows it to use content from across the web, including from publishers who had opted out via Google-Extended, for features like AI Overviews.

AI Overviews are Google’s AI-generated summaries that appear directly in search results, aiming to answer user queries instantly. But for publishers, these summaries have sparked concerns. They fear that users no longer need to click through to websites, leading to what’s being called the “zero-click” problem. Some publishers estimate traffic drops of 20% to 60%, and the publishing platform Raptive believes this could result in up to $2 billion in annual ad revenue losses across the industry.

The dilemma is especially tough for smaller websites, as they have already lost over half their traffic since AI Overviews rolled out.

For publishers, the reality is harsh: Google controls over 90% of the search engine market. Not showing up in Google Search is simply not an option. And so, many are being forced to accept that their content will be used to train the very tools that are threatening their traffic and revenue.

Google, for its part, says publishers still have control over how their content appears in search and that AI models have long been a part of the search experience. The company also downplayed the internal memo as an early draft that didn’t represent a final decision.

However, testimony from Google DeepMind’s VP Eli Collins confirmed that even publishers who opted out of AI training still see their content used in AI Overviews—deepening worries about transparency and consent.

This issue is now front and centre in Google’s antitrust case, where Judge Amit Mehta is reviewing potential remedies to reduce Google’s market dominance. One proposal is to let publishers opt out of AI Overviews without affecting their search rankings—a move Google has so far resisted.

For now, publishers remain stuck between a rock and a hard place—choosing between losing control over their content or losing visibility altogether.