The third Hungarian reference against Google over its chatbot and press publications asks the Court of Justice eleven questions, from whether training a model is part of an 'information society service' to whether a robots.txt file is an adequate way of opting out of text and data mining. It reaches the Court two weeks after Advocate General Szpunar proposed that training falls outside the press publishers' right.
The second Hungarian reference against Google on chatbots and the press publishers' right goes further than the first: it asks about summaries rather than copied passages, about grounding on current articles, and whether a publisher that did not opt out has consented. It arrives on the day Advocate General Szpunar proposed that the Court leave training aside in Like Company.
A Hungarian publisher says Google's chatbot reproduced its articles — once when the model was trained, again when it answered. Google says a model that predicts the next word copies nothing. The first reference on generative AI and copyright asks the Court which of them is right.
On 14 June 2017 the Court ruled (ECLI:EU:C:2017:456) that the operators of The Pirate Bay themselves make a communication to the public by indexing and managing torrents of protected works, despite the content sitting on users' machines.
On 31 May 2016 the Grand Chamber ruled (ECLI:EU:C:2016:379) that the operator of a rehabilitation centre showing TV broadcasts on screens in its waiting and training rooms communicates works to the public and must pay the collecting societies — aligning…
On 8 September 2016 the Court ruled (ECLI:EU:C:2016:644) that posting a hyperlink to works published online without the rightholder's consent is a communication to the public where the poster knew or ought to have known of the illegality.