International Press Release

Like Company’s Statement regarding the AG Opinion dated 3 November 2026 in the CJEU case C-250/25 (Like Company v. Google) 

Like Company views the AG Opinion’s findings as a firm step towards securing fair compensation for high-quality content that fuels the generative AI revolution. The Opinion confirms that the AI industry and generally machine intelligence cannot unilaterally extract value from European journalism and human creativity without accountability.

The Like Company v. Google copyright case extends far beyond Google: a favourable CJEU ruling will set a binding landmark precedent across the European Union, establishing a clear legal basis for press publishers to pursue claims also against OpenAI, Anthropic, Meta, and other AI service providers for unauthorised AI training and outputs. Beyond copyright, this case will be recognised as one of the foundational pillars forging the Legal Principle of Human Primacy.

Two distinct rights – two separate remunerations

Like Company claims that Article 15 of the CDSM Directive enacts two distinct and standalone rights: the right of reproduction and the right of communication to the public. These rights serve different purposes and must be recognised – and compensated – independently.

We strongly welcome the AG Opinion’s conclusion that the display of AI-generated answers can infringe both of these rights. This means that the compensation due for online publishers must reflect both types of uses, the training of AI models and the chatbots. The Opinion makes it clear that AI service providers cannot hide behind technological complexity to avoid compensating publishers and AI shall not diminish or replace the value of human creativity.

No automatic consent

Furthermore, we are highly satisfied that the Opinion adopted our stance on the necessity of publisher consent regarding these AI uses and fully agreed with our arguments raised at the Luxembourg hearing earlier this year. As the Advocate General emphasized, implied consent cannot possibly arise without publishers being actually and individually informed in advance. Lacking this essential transparency, Google cannot evade liability by pointing to technical measures that initially also failed to provide an AI-specific opt-out.

The Opinion also aligned with our arguments that the text and data mining (TDM) exception involves certain legal thresholds to balance interests. The TDM exception is not a free pass for verbatim copying. For example, it cannot justify instances where an AI chatbot like Gemini reproduces protected content word-for-word in its answers.  At the oral hearing of the case in response to a judge’s question, Google’s lawyers from international law firm Hengeler Mueller expressly admitted that the possibility of Gemini producing a verbatim copy of protected content cannot be excluded.

Not just a “Dolphin Article”, but a longer period of extensive AI training is indeed part of the case

We must respectfully note that the Opinion reflects a factually inaccurate narrative heavily promoted by Google in the press: that this dispute concerns only a single article about dolphins and that AI training is “hypothetical”.

Google’s ongoing efforts to reframe the case as a minor dispute over one article is simply a strategy to avoid paying for the vast amounts of data used to train its AI models. The reality confirming the contrary is in the court files: the Hungarian national court explicitly established and confirmed in a formal court order upon the request of the CJEU before the oral hearing of the case in Luxembourg that the factual scope of the national litigation covers Google’s continuous use of multiple Like Company publications for both training the Gemini models and generating outputs over a longer time period. AI training is a legally established fact in this case, not a hypothesis.

Nevertheless, we highly value the Advocate General’s meticulous legal reasoning regarding AI training. We trust that the CJEU will rule on the merits of this issue taking into account that the scope of the national litigation is much wider than a single article. Following the Advocate General’s substantive reasoning, we are confident that the CJEU’s judgment will find that AI training in itself can also constitute a standalone infringement of the right of reproduction – which, in and of itself, entitles the publishers to fair compensation, regardless of the infringements connected to AI outputs.

Budapest, 4 November 2026