EU Court Adviser Warns Google Gemini May Violate Copyright Rules When Reproducing Press Articles and Bypassing Paywalls

The legal landscape surrounding artificial intelligence and digital publishing faces a potential watershed moment following a significant opinion issued by an adviser to the European Union’s highest court. Maciej Szpunar, Advocate General at the Court of Justice of the European Union (CJEU), declared that Google’s AI-powered chatbot, Gemini, may infringe upon the copyright protections of press publishers when it reproduces recognizable portions of their work or unlawfully bypasses paywalls and other technical access barriers. The legal entanglement, originating from a dispute between Hungarian website operator Like Company and tech conglomerate Google, highlights the escalating tension between creators of original journalism and the developers of generative AI models whose platforms increasingly capture online audiences before they ever click through to a publisher’s primary website.

The outcome of this high-stakes legal battle, expected in the coming months, carries profound economic and structural implications for the digital media ecosystem across Europe and beyond. Traditional publishing houses have watched helplessly as referral traffic from search engines plummets, replaced wholesale by direct, conversational AI summaries that synthesize web content without sending readers to the underlying source. If the CJEU ultimately adopts the Advocate General’s reasoning and rules against Google, the decision could establish a legally binding precedent requiring tech giants to compensate publishers for utilizing their journalistic output, offering a vital economic lifeline to an industry struggling to monetize its digital footprint in an automated era.

The Genesis of the Dispute: Like Company v. Google

The legal confrontation currently before the Luxembourg-based court began as a David-and-Goliath struggle in Hungary, spearheaded by the digital media operator Like Company. Like Company initiated legal action against Google, alleging that the American multinational technology corporation infringed upon its intellectual property rights by feeding, retrieving, and reprocessing its published journalistic content through the Gemini artificial intelligence ecosystem to directly answer user queries.

For years, search engine optimization (SEO) and referral clicks served as the foundational economic engine for digital publishers. Users searching for current events or specialized information would enter queries into search engines, receive a list of relevant hyperlinks, and click through to the publisher’s website, where advertisements or subscription paywalls generated revenue. However, the paradigm shifted dramatically with the widespread integration of generative artificial intelligence into search platforms. Rather than directing users to external URLs, conversational agents like Gemini now aggregate, synthesize, and present the core findings of news articles directly on the search results page.

Like Company argued before the courts that this practice strips away the economic value of hard-earned journalism, effectively siphoning away readership and advertising revenue while exploiting copyrighted expression without authorization or financial compensation. The case eventually escalated to the Court of Justice of the European Union, where judges sought the formal legal opinion of Advocate General Maciej Szpunar to clarify how existing European copyright law applies to modern large language models.

Legal Boundaries: Training Versus Display

In his detailed legal opinion, Advocate General Szpunar drew a critical distinction between how large language models are initially trained and how they subsequently retrieve, process, and display information to end users. This dichotomy forms the core of the adviser’s legal framework and provides a nuanced view of copyright compliance in the age of generative AI.

According to Szpunar, the use of copyrighted material to train foundational LLMs—which ingest massive quantities of text to recognize patterns, understand human language, and generate contextual responses—is "freely permitted under the text and data mining exception" established within European Union legal frameworks. The advocate general noted that within the specific parameters of the case at hand, there was no concrete evidence demonstrating that Google had unlawfully acquired or utilized Like Company’s specific articles for the direct training of the Gemini model.

However, the legal permissibility shifts drastically once the model transitions from the training phase to the active retrieval and generation phase. Szpunar asserted that when an LLM-driven chatbot actively retrieves a publication from a press publisher’s website, reuses that publication to formulate a direct answer, and presents that synthesized information to a user, the rights of the press publisher may indeed be infringed under specific circumstances.

Specifically, the advocate general identified two primary triggers for copyright infringement during the generation phase:

1. Circumvention of Technical Barriers: When chatbots access publications by unlawfully bypassing paywalls, login screens, or other technical protection measures designed to restrict access.
2. Recognizable Reproduction of Expression: When the AI reuses the specific "form of expression" of a publication in a recognizable manner, effectively echoing the distinctive prose, creative structure, or unique phrasing crafted by human journalists.

Szpunar took care to reiterate a fundamental tenet of international and European copyright doctrine: copyright law protects the specific form of expression chosen by an author, but it does not protect mere underlying facts. Therefore, while an AI chatbot summarizing pure historical data or basic, unvarnished facts reported by multiple outlets may remain legally permissible, reproducing the distinct creative output, analytical framework, or stylistic expression of a specific press publication crosses the legal threshold into infringement.

Industry Reactions and Economic Implications

The legal opinion has sent shockwaves through both the technology sector and the international publishing community, eliciting strong responses from industry associations and digital rights advocates. Media organizations have long maintained that major search and AI conglomerates abuse their dominant market positions to extract valuable journalism for free, starving newsrooms of the revenue necessary to sustain investigative reporting.

Angela Mills Wade, executive director of the European Publishers Council (EPC), emphasized the severe power asymmetry inherent in the modern digital marketplace. "Google’s powerful gatekeeping role in search makes this particularly important. Publishers depend on Google to reach readers, while Google increasingly delivers AI-generated answers directly to users," Mills Wade stated following the release of the adviser’s opinion.

Mills Wade further criticized the difficult ultimatum facing modern media outlets, noting that publishers should never be forced to choose between maintaining basic search visibility and protecting their fundamental intellectual property rights. Furthermore, she argued that Google’s absolute control over access to digital audiences must not be weaponized as leverage to secure high-value journalistic content for proprietary AI services without fair compensation.

Representatives for Google, meanwhile, have defended their approach to AI development and information retrieval, arguing that their systems empower users to discover diverse sources of information across the web while driving meaningful engagement and traffic to publishers. Tech sector advocates frequently argue that restrictive legal interpretations could stifle European technological innovation, handicap the deployment of generative AI tools within the single market, and isolate European enterprises from global competitors.

The Broader Context: Global Regulatory Pressure on AI

The proceedings before the Court of Justice of the European Union do not occur in a vacuum. Across the globe, lawmakers, regulatory bodies, and courts are grappling with the unprecedented challenges posed by generative artificial intelligence. The European Union has positioned itself as a global regulatory pioneer in the digital sphere, implementing comprehensive legislative frameworks such as the EU Artificial Intelligence Act and the Digital Markets Act, alongside the EU Copyright Directive, which established specific neighboring rights for press publishers.

As these legal frameworks mature, tech giants find themselves facing mounting multi-front litigation. Numerous copyright infringement lawsuits have been filed in the United States and Europe by authors, musicians, visual artists, and news publishers against major AI developers, including OpenAI, Microsoft, Meta, and Google. These plaintiffs universally challenge the unauthorized harvesting of copyrighted works to train and operate commercial artificial intelligence systems.

While Advocate General Szpunar’s opinion is strictly non-binding, historical precedent indicates that the Court of Justice of the European Union follows the majority of its advisers’ opinions when issuing final judgments. If the court formally adopts Szpunar’s legal reasoning in the coming months, it will establish a powerful pan-European standard. Such a ruling would compel tech conglomerates to fundamentally redesign how generative AI products interact with web-based journalism, potentially forcing them to negotiate licensing agreements, implement strict content-filtering mechanisms, or risk substantial financial liabilities for copyright infringement.

Conclusion and Outlook

The looming decision by the Court of Justice of the European Union represents a critical milestone in defining the legal boundaries of the digital age. As artificial intelligence systems rapidly evolve from experimental novelties into the primary infrastructure of the internet, the fundamental question of who owns, controls, and profits from human creativity remains urgently unresolved.

For European publishers fighting to preserve their economic viability in an era of conversational search engines and zero-click information retrieval, the advocate general’s opinion offers a glimmer of hope that the law will adapt to protect original journalism. For tech giants like Google, the opinion serves as a stark warning that the era of unfettered content extraction may be drawing to a close. As the court prepares to deliver its final binding ruling in the months ahead, the digital media and technology sectors alike wait with bated breath for a verdict that will shape the future of information sharing for decades to come.