Lower traffic and less revenues to publishers

Trade associations representing publishers' interests claim that AI overviews cause a drop of up to 25 % in publishers' traffic, or even up to 50 % according to a digital marketing company. A network of digital marketing companies reports that 58 % of Google searches now end without the user visiting a single link. Reduced traffic translates into fewer ad impressions and a subsequent loss of revenue for content creators. In response, a coalition of publishers has filed a formal antitrust complaint with the European Commission against Google, accusing its 'AI Overviews' of diverting traffic and revenue from them. An alliance of German media and digital industry NGOs, associations and organisations has filed a formal complaint in Germany. A third case launched in the UK confirms Google has strategic market status in search services, including AI overviews and AI mode, the latter being its AI chatbot launched to compete against other chatbots.

The plaintiffs believe AI overviews are increasingly monetised through integrated advertisements, while Google extracts content from third-party websites without offering direct compensation or opt-out options, and without obtaining prior consent. They claim this has implications for media diversity, freedom of opinion and democratic discourse. Representatives of publishers' interests claim they cannot opt out from appearing in the overviews without also being suppressed from the regular search listing.

Difficulties in the training of AI models

Tools like Google's AI Overviews rely on two components: a large-language model (LLM) and a system able to retrieve information. Both need data, the first to learn to use human language and the second to retrieve documented information to answer users' requests. Users and businesses may have grown accustomed to their data being collected and used for search results or targeted advertising, but the same may not be true for AI training. Complaints in the United States (US) point to an update of Google's privacy policy, adding that the company may use publicly accessible information to train its AI tools. In fact, training LLMs with data publicly available on the web may raise copyright issues. Similarly, there is a case against Meta for not respecting the EU General Data Protection Regulation (GDPR) when collecting data to train their model, as users were not asked for prior consent to their data being collected for this purpose. However, recently proposed amendments to the GDPR might clarify that processing personal data for AI training may be permissible. Stakeholders are challenging Google. For instance, Cloudfare, a global platform and network of data centres that maintain websites and applications, is supporting a consortium of publishers to collectively block AI crawlers from collecting content for training unless companies pay for the content.

Threat to the open web

Until recently, Google drove online traffic in the open web, the public, interconnected network of websites and online services used for browsing, shopping and consuming news. Thus, Google has helped users find information by crawling and indexing web pages and ranking them based on a number of criteria. This is changing now, as users are offered generative AI-based responses and conversations and increasingly stop clicking and driving online traffic. Google has admitted that the open web is in rapid decline, referring to open-web display advertising. However, many believe this decline is much broader, as AI-driven search poses an unprecedented threat to the open web. In addition, users are increasingly using chatbots to conduct online searches, further reducing the number of users who access websites from search engines.

Risks for information quality, critical thinking and digital inclusion

AI-generated searches may include incorrect or biased information. Research suggests that AI could produce 'chat chambers' that reinforce the misinformation it hallucinates. Thus, 27 % of US adults do not trust AI-generated search results. Concerns around AI-generated misinformation are high, with the likelihood of chatbots repeating false information nearly doubling from 18 % in August 2024 to 35 % in August 2025. In addition, higher confidence in generative AI is associated with less critical thinking, according to research. It could also widen the digital divide, as young users are more AI-literate than older ones, but young students are increasingly dependent on AI use in classwork, risking lower creative thinking.