Google's AI Search Problem: Why Regulators Are Stepping In

Google is facing regulatory investigations in Britain and Europe over how it uses news articles and web content to power its new AI search features. The core complaint is straightforward: Google is using publishers' content to train AI systems and generate search answers without letting those publishers opt out—or giving them any control over the process.

The UK's Competition and Markets Authority wants Google to offer a middle ground. Under their proposal, news sites and publishers could say "don't use our content for AI" while still appearing in regular Google search results. Right now, publishers face an all-or-nothing choice: either let Google use their content for AI features, or disappear from search entirely.

Europe's competition authorities opened a similar investigation. They're asking whether Google's approach is fair—specifically, whether it's reasonable to use content from news publishers and YouTube creators without their permission, without paying them, and without a way to opt out.

How Google's AI Search Actually Works

When you use Google's new AI search features, you get answers written by AI that summarize information from multiple web pages. These features require users to join a waiting list and use a special tool called Search Generative Experience.

Google treats content that appears in these AI answers the same way it tracks traffic in regular search results. The company says publishers should use standard search optimization techniques for both types of search.

The problem regulators see is this: Google is bundling two different things together. Publishers cannot pick and choose. They either accept their content being used in AI systems, or they lose visibility in Google search altogether. Internal documents show Google made a firm decision that web publishers should have to participate.

This bundling extends to YouTube. The European investigation will look at whether YouTube video creators got proper notice that their videos could be used to train Google's AI systems.

What This Means for News Sites and Publishers

AI search answers create a dilemma for publishers. When Google's AI summarizes multiple news articles into a single answer on the search results page, users may read the summary and leave—they don't visit the original news site. This can reduce traffic and advertising revenue.

Publishers are caught in a difficult position. Their articles help train the very AI systems that may send fewer readers their way. And they have no way to say "use our content for regular search, but not for AI training."

This isn't the first time publishers have faced this kind of challenge from tech platforms. When smartphones became dominant, news sites had to adapt to mobile-first search, and Google could control how their traffic flowed. But AI search is different in a fundamental way. Rather than just changing how people find articles, it can answer questions without anyone visiting the publisher's site at all.

What Regulators in Multiple Countries Are Doing

Britain and Europe are both investigating Google's AI search, and they seem to be coordinating their efforts. The European investigation covers multiple parts of Google's AI work: where it gets training data, how it affects competition, and what it means for other search engines.

Britain is using a new law called the Digital Markets, Competition and Consumers Act to set rules. Both countries want the same basic things: let publishers control their own content and make sure creators are compensated or at least have a choice.

What's notable is that regulators are treating AI features as something distinct from regular search. They understand that training AI models and generating search answers are different uses of content than the traditional way Google indexes and ranks web pages. That's a sophisticated way to think about the problem.

The Difficult Technical Questions Ahead

If regulators force changes, Google will need to build new systems. Creating an opt-out tool means building technology that can distinguish between traditional search indexing—where Google finds and ranks your article—and AI training, where it uses your words to teach its AI system.

Publishers would need simple tools to set their preferences. Google's systems would have to respect those choices across its entire AI pipeline. That's more complex than it sounds.

There's also a timing problem. Much of the AI training data already includes content from publishers who never got asked for permission. Fixing this might mean retraining Google's AI with only approved content, or filtering search results to exclude certain publishers.

The European investigation mentions another idea: letting competing search engines see some of Google's data about popular searches and how sites rank. That adds even more technical challenges around privacy and fair competition.

The broader context here is that regulators are still learning how to govern AI systems in large platforms. These investigations will likely set precedents that affect how other tech companies—not just Google—build AI features. The outcomes matter beyond just news publishers. They could establish rules about when companies need permission to use content for AI training and whether content creators deserve compensation or control.

Google continues adding new capabilities to its search AI as these investigations proceed. That ongoing evolution could make it harder to reach a final resolution.