EU vs. Gemini: The legal clash redefining AI and copyright

Google's legal crossroads in Europe: The end of the extractive content era?

The generative artificial intelligence ecosystem has entered a phase of forced regulatory maturity. The recent opinion of the Advocate General of the Court of Justice of the European Union (CJEU), Maciej Szpunar, regarding the conflict between the Hungarian operator Like Company and the tech giant Google, is not merely a procedural step; it is a milestone that marks the definitive boundary between technological innovation and intellectual property rights in the digital age. Although the opinions of the Advocates General are not binding, the CJEU follows their recommendation in the vast majority of cases, making this ruling a stern warning for developers of large language models (LLMs).

The friction point: The distinction between training and exploitation

Szpunar's central thesis establishes a critical technical and legal distinction that could reshape the AI market. On one hand, the Advocate General reaffirms that model training through text and data mining (TDM) techniques is protected under current European regulatory exceptions. This is a partial victory for the tech industry, which argued that restricting model training would amount to a technological blockade against U.S. and Chinese competitors.

However, the breaking point occurs in the inference or response generation phase. Szpunar is emphatic: when a chatbot like Gemini accesses protected publications—especially if it uses methods to bypass paywalls or technical protection measures—and replicates the "specific form of expression" of the author, it commits a direct copyright infringement. This analysis shifts the focus from the machine learning process to the final product received by the user, suggesting that AI cannot be a refuge of impunity under the guise of being a "summary" or "index" system.

Impact on the media industry and the end of organic traffic

The journalistic industry has lived for decades under the dependency of Google's search algorithm. Historically, the tacit agreement was clear: Google indexed content in exchange for sending referral traffic via links. With the transition toward the Search Generative Experience (SGE) and direct answers, this business model has fractured. Angela Mills Wade, executive director of the European Publishers Council, has pointed out an uncomfortable truth: Google uses its position as a gatekeeper of audience access to extract value without compensating the original creators.

This conflict is reminiscent of the "Google Tax" disputes of the last decade in countries like Spain or Germany, where attempts were made to charge aggregators for the use of snippets. The current difference lies in the scale and nature of the content: it is no longer about a simple link, but a synthetic response that eliminates the user's need to visit the original website. If the CJEU ratifies this position, Google could be forced to establish a massive licensing system, similar to the one already implemented in markets where regulation mandates compensating publishers for the use of their content in aggregators.

What to expect from this ruling and its precedents?

• Training vs. Response: The ruling creates a legal dichotomy. Data mining for technological development is lawful, but the commercial exploitation of that information in the final product requires authorization and, potentially, remuneration.
• The end of technical circumvention: Szpunar's opinion specifically punishes the evasion of paywalls. This means that AI companies cannot justify access to premium content under the premise of "public access" if the publisher has implemented technical barriers, setting a precedent against aggressive scraping techniques.
• Economic reconfiguration: If Google is held liable, the "free AI" model becomes unsustainable without a robust licensing structure. This could lead to greater fragmentation: publishers who proactively block AI bots (via robots.txt or contractual agreements) versus those who decide to negotiate fees for appearing in Gemini's summaries.

In perspective, we are facing an event comparable to the implementation of the GDPR (General Data Protection Regulation) in 2018. That moment forced companies to treat personal data as an asset under audit; this new ruling will force AI companies to treat protected content as an asset under license. Speculation about a possible fine or imposition of fees is high, but what is certain is that the CJEU is aligning European digital sovereignty with the protection of its creative industries. The era of "everything for free" to train and feed AIs has ended, forcing tech giants to move toward a model of transparency and compensation that, while more costly, is the only path toward legal sustainability in the most regulated market in the world.