{"schema":"sanctity-signal/0.2","record_type":"signal","id":"cjeu-ag-genai-copyright-e072","domain":"governance","drop_id":"signal-e072","occurred_at":"2026-11-03","how_to_use":"This is a structured record of one real-world action and its competing interpretations, from Sanctity Signal. To build content from it (a reel script, a blog, a thread): open on the hook, ground the viewer with plain and observed, walk the dated trail, present the interpretations as live tension between named readings, and close on the before/after shift and what_to_watch. Hard rules: items under established are fact; items under unknown are never presented as fact; the evidence status (verified or reported) is stated, not dropped; sources are cited.","actor":"Court of Justice of the European Union","action_type":"delivery of an Advocate General's opinion in the first generative AI copyright case to reach the European Union's highest court","observed":"Case C-250/25, Like Company Kft. v Google Ireland Limited, is a reference from the Budapest Regional Court in Hungary. Like Company runs advertising-funded Hungarian news portals. Its disputed article, published July 21, 2023, concerned a plan to bring dolphins to Lake Balaton; Gemini's summary invented a pool 25 metres long holding 1,000 cubic metres. The claim period runs June 13, 2023 to May 31, 2024. The March 10, 2026 hearing drew observations from both parties, the European Commission and 9 governments. Advocate General Maciej Szpunar delivered his opinion on November 3, 2026: 240 numbered points, 243 footnotes, far longer than usual, more than 90 points on training alone. The Hungarian court referred 4 questions, 2 on training and 2 on answers. On training, Szpunar advises declaring the questions inadmissible: the article predates the models' training, and training is not online use under Article 15(1) of Directive 2019/790. Answering anyway: training does involve reproductions under Article 2 of the InfoSoc Directive, and the text and data mining exception in Article 4(1) in principle applies, conditional on lawful access and machine-readable opt-outs under Article 4(3). He names robots.txt and Google-Extended, released September 28, 2023, as qualifying, and writes that a trained model is not a database. Where a model routinely regurgitates, the exception should be refused absent mitigation measures. On answers: retrieval is online use, copying occurs through it, and protection covers the form of expression, not facts. The test is whether the answer substitutes directly, so the reader need not visit the original. Circumventing a paywall or bot block infringes. Making available to the public needs recognisability amounting to substitution, which he calls rather unlikely given non-deterministic outputs, though possible through negligent retrieval design. He rejects Google's implied-consent argument, citing VG Bild-Kunst, and proposes a presumption shifting the burden to the provider once a rightholder shows a reasonable sample of resembling answers; one example rarely suffices. On the facts he is unfavourable to Like Company: a purely informative report, no technical measure apparently circumvented, Google apparently mitigating. Judgment comes at a date not set. Not established: how the Court will rule, and whether Like Company's portals are press publishers.","evidence":{"status":"reported","note":"Reported, not verified. The Court's own documents refused every fetch today: press release 155/26 at curia.europa.eu and the opinion on InfoCuria both failed. The account rests on PPC Land, which quotes the opinion at length, with mLex and Courthouse News. Verification event: the Court publishes the opinion text, and later its judgment.","sources":[{"role":"primary","label":"Court of Justice press release 155/26 and the opinion in C-250/25, Nov 3, 2026 (both refused every fetch)","url":"https://curia.europa.eu/site/upload/docs/application/pdf/2026-11/cp260155en.pdf"},{"role":"corrob","label":"PPC Land: Google's Gemini can infringe press rights in 2 ways, EU court adviser says, Nov 3, 2026","url":"https://ppc.land/googles-gemini-can-infringe-press-rights-in-2-ways-eu-court-adviser-says/"},{"role":"corrob","label":"Courthouse News: EU legal adviser finds limited uses of Google AI could break publishers' rights","url":"https://www.courthousenews.com/eu-legal-adviser-finds-limited-uses-of-google-ai-could-run-afoul-of-publishers-rights/"},{"role":"corrob","label":"mLex on the Advocate General's opinion in C-250/25, Nov 3, 2026","url":"https://www.mlex.com/mlex/artificial-intelligence/articles/2545531"},{"role":"corrob","label":"InfoCuria case file for C-250/25, Like Company v Google","url":"https://infocuria.curia.europa.eu/tabs/jurisprudence?sort=DOC_DATE-DESC&searchTerm=%22C-250%2F25%22&publishedId=C-250%2F25"}]},"strength":{"cost":"Low","commitment":"Medium","reversibility":"Reversible","persistence":"Years","rationale":"Cost is low: an opinion is a document, and its author risks nothing material. Commitment is medium, because 240 reasoned points on the record constrain the argument even where the Court departs from them. Reversibility is high: the Court can disagree outright, and only the judgment binds. Persistence is years, because a ruling on training and output liability governs European AI copyright long after this dispute ends. Scored as a strong indicator, not a decided fact."},"shift":{"prior":"European courts had not yet tested whether AI training infringes copyright.","action":"Case C-250/25, Like Company Kft. v Google Ireland Limited, is a reference from the Budapest Regional Court in Hungary. Like Company runs advertising-funded Hungarian news portals. Its disputed article, published July 21, 2023, concerned a plan to bring dolphins to Lake Balaton; Gemini's summary invented a pool 25 metres long holding 1,000 cubic metres. The claim period runs June 13, 2023 to May 31, 2024. The March 10, 2026 hearing drew observations from both parties, the European Commission and 9 governments. Advocate General Maciej Szpunar delivered his opinion on November 3, 2026: 240 numbered points, 243 footnotes, far longer than usual, more than 90 points on training alone. The Hungarian court referred 4 questions, 2 on training and 2 on answers. On training, Szpunar advises declaring the questions inadmissible: the article predates the models' training, and training is not online use under Article 15(1) of Directive 2019/790. Answering anyway: training does involve reproductions under Article 2 of the InfoSoc Directive, and the text and data mining exception in Article 4(1) in principle applies, conditional on lawful access and machine-readable opt-outs under Article 4(3). He names robots.txt and Google-Extended, released September 28, 2023, as qualifying, and writes that a trained model is not a database. Where a model routinely regurgitates, the exception should be refused absent mitigation measures. On answers: retrieval is online use, copying occurs through it, and protection covers the form of expression, not facts. The test is whether the answer substitutes directly, so the reader need not visit the original. Circumventing a paywall or bot block infringes. Making available to the public needs recognisability amounting to substitution, which he calls rather unlikely given non-deterministic outputs, though possible through negligent retrieval design. He rejects Google's implied-consent argument, citing VG Bild-Kunst, and proposes a presumption shifting the burden to the provider once a rightholder shows a reasonable sample of resembling answers; one example rarely suffices. On the facts he is unfavourable to Like Company: a purely informative report, no technical measure apparently circumvented, Google apparently mitigating. Judgment comes at a date not set. Not established: how the Court will rule, and whether Like Company's portals are press publishers.","update":"The test is lawful access and whether an answer replaces the article."},"inference":{"established":["An Advocate General delivered an opinion in Case C-250/25 on November 3.","The opinion is advisory. The Court has set no judgment date.","It is the first generative AI copyright case to reach the Court."],"unknown":["That the Court will agree. Opinions are followed often, not always.","Any outcome for Like Company. The Hungarian court decides the facts.","That training is cleared. The exception needs lawful access and opt-outs."]},"narrative":{"hook":"The EU's top court was advised that AI training is mostly lawful but answers can infringe.","plain":"On November 3 an Advocate General told the Court of Justice that model training in principle falls inside Europe's text and data mining exception. Caveat: the opinion is advice, not a ruling.","story":["Advocate General opinions guide the Court, which often follows them. If it does, the fight moves off training data and onto two narrower questions: did the model get past a paywall or a robots file, and does its answer substitute for the article? Publishers would sue over outputs, not corpora."],"trail":[{"d":"JUN 2023","t":"Like Company publishes the disputed article on July 21, inside a claim period opening June 13. Google releases Google-Extended, the crawler opt-out the opinion later treats as valid, on September 28, and Gemini 1.0 in December."},{"d":"MAR 10","t":"The Court hears Case C-250/25. The Commission and 9 governments file observations, an unusually crowded hearing for a reference from one Hungarian regional court."},{"d":"OCT 30","t":"Anthropic files an internet specification for a single machine-readable AI training opt-out signal. Edition 067. The opinion delivered 4 days later makes machine-readable opt-outs the condition on which the training exception turns."},{"d":"OCT 30","t":"Britain's competition regulator opens an abuse case over Meta's terms for publishers' content. Edition 067. Publishers press the same grievance through competition law and copyright at once."},{"d":"NOV 3","t":"Advocate General Szpunar delivers a 240-point opinion: training in principle inside the text and data mining exception, chatbot answers infringing where access was unlawful or the answer substitutes for the article. Advisory only."},{"d":"Reacts","r":true,"t":"The Court's judges, deliberating with no judgment date; the Budapest Regional Court; Google, whose Gemini grounding is the subject; European publishers, including the 5 French groups that complained October 29; OpenAI and Anthropic, who rely on the same exception; the Commission, which intervened; every rightholder choosing between an output case and a training case."}],"glossary":[{"term":"Advocate General","def":"An independent adviser at the Court who publishes a reasoned opinion before judgment. It binds nobody, and the Court follows it in most cases, not all."},{"term":"Text and data mining exception","def":"A European rule letting anyone analyse lawfully accessible works without permission, unless the rightholder reserved the use in a machine-readable way."},{"term":"Reference for a preliminary ruling","def":"A national court asks what European law means. The answer binds across the Union; that court still decides its own case."},{"term":"Retrieval-augmented generation","def":"The model is handed a document's text at question time rather than relying on training. The opinion treats this retrieval as online use."}]},"judgment":{"prompt":"If the Court follows: which explanation do you believe?","interpretations":[{"id":"a","label":"The law settled on outputs","text":"Training cleared in principle; infringement moves to answers that replace the article.","abstain":false},{"id":"b","label":"One adviser, not the Court","text":"An Advocate General's view in one Hungarian case about one 2023 article.","abstain":false},{"id":"c","label":"A deliberate narrowing","text":"The opinion hands publishers an evidence presumption while sparing the training corpus.","abstain":false},{"id":"d","label":"Normal EU process","text":"Every big reference gets an opinion, then a judgment months later.","abstain":false},{"id":"e","label":"Too early to call.","text":"The evidence cannot yet separate these.","abstain":true}],"distribution":{"status":"held back until you have made your call","calls":null}},"story_thread":null,"what_to_watch":[{"linked_interpretation":"a","direction":"strengthens","observable_event":"The Court gives judgment in C-250/25 holding that training falls within the text and data mining exception, while output liability turns on lawful access or substitution, by December 31, 2027.","window_end":"2027-12-31","state":"unresolved"},{"linked_interpretation":"a","direction":"weakens","observable_event":"The Court holds that training requires a licence, or declares all 4 questions inadmissible without ruling on the exception, by December 31, 2027.","window_end":"2027-12-31","state":"unresolved"},{"linked_interpretation":"c","direction":"strengthens","observable_event":"A European court adopts the proposed presumption, shifting the burden to a provider once a rightholder shows a sample of resembling answers, by June 30, 2028.","window_end":"2028-06-30","state":"unresolved"}]}