EU court adviser warns Google Gemini could violate press copyright

The legal distinction of AI

An adviser to the Court of Justice of the European Union (CJEU) has provided a significant legal opinion on how artificial intelligence interacts with copyright laws. Advocate General Maciej Szpunar highlighted a major distinction between the two primary stages of AI development: training and inference. While the two processes are linked, the legal protections surrounding them differ significantly under European law.

According to the court statement, the training of large language models (LLMs) is generally considered permissible. These systems use vast amounts of text to learn how to generate human-like responses and summarize information. Because this process is viewed as a form of learning, it typically benefits from existing text-and-data-mining (TDM) exceptions that allow for the use of copyrighted material for analysis.

The legal landscape shifts when the AI moves from learning to providing answers to a human user. The opinion suggests that when a chatbot like Google's Gemini retrieves a publication and reuses it to formulate a response, it may cross a legal line. This distinction is crucial for tech giants who currently rely on the broad permissions granted for data mining to build their products.

How copyright is infringed

The core of the legal concern lies in the specific way an AI chatbot presents information to a user. Szpunar noted that copyright laws do not shield the basic information or data points shared by news outlets. Instead, these laws are designed to protect the unique way a writer or journalist chooses to describe those facts.

If an AI chatbot creates dialogue that mirrors protected news material, specific excerpts, or unique phrasing without permission, it risks violating EU rules. The court adviser stated that a press publisher's rights may be violated in certain situations when the AI uses a publication to build an answer. This is particularly relevant when the AI provides a version of the original writing that is easily identified.

Furthermore, the opinion points out that legal issues arise when chatbots gain access to publications by unlawfully circumventing paywalls or other technical barriers. When a chatbot bypasses these protections to pull content and then presents it to a user, it enters a high-risk zone for copyright infringement. This creates a complex hurdle for developers who must balance accessibility with the legal rights of content creators.

Impact on digital publishers

The potential ruling could provide a significant lifeline to online media companies that are currently struggling with declining traffic. Many publishers have observed a drop in visitors as AI-generated summaries begin to replace traditional search engine results. Because these summaries often provide the answer directly, users may no longer click through to the original source via a hyperlink.

The opinion suggests that a ruling against Google could force the tech giant to pay for the use of their content. This would establish a new economic model where AI companies must compensate the creators whose work they use to inform their responses. For many digital publishers, this could be the difference between remaining viable and facing financial collapse in an AI-driven era.

As The News International reported, the situation creates a tension between the convenience of AI tools and the sustainability of journalism. If AI chatbots continue to replace the traditional search-and-click model without compensation, the ecosystem of news production could be fundamentally altered. The court's opinion serves as a warning that the current status quo may not be legally sustainable for tech platforms.

The road to a final ruling

While the opinion delivered by Advocate General Szpunar is non-binding, it carries immense weight for the future of the industry. It serves as a primary guide for the Court of Justice of the European Union as it prepares for a definitive ruling. This final decision will set a precedent for how generative AI is deployed across all 27 member states of the European Union.

The court will need to determine the exact boundaries between permissible data mining and prohibited reproduction of content. This decision will impact not only Google's Gemini but also all other tech platforms and generative AI developers. The outcome will determine whether these companies must fundamentally rethink their current methods for handling journalism and digital publisher rights.

For the broader tech industry, the ruling could lead to a standardized framework for licensing content for AI use. It may also lead to more sophisticated technical barriers to prevent AI from scraping content in ways that violate copyright. The legal community is watching closely to see how the court balances the innovation of AI with the protection of intellectual property.