Summary: On March 10, 2026, the European Parliament approved an own-initiative report on copyright and generative AI by 460 votes to 71, with 88 abstentions. It presumes commercial AI developers used copyrighted works unless rebutted and calls for explicit consent over press publishers’ training uses. A related Poynter article describes statutory licensing proposals that would require AI companies to pay for journalism, while noting differing national approaches.
The European Parliament’s plenary voted through the own-initiative report “Copyright and generative artificial intelligence — opportunities and challenges” (procedure 2025/2058(INI), with Axel Voss as rapporteur) on March 10, 2026, finishing at 460 in favor, 71 against and 88 abstentions. That same week, on March 9, Poynter ran an advocacy article by Columbia University’s Anya Schiffrin and the European University Institute’s Roberta Carlini pressing for “statutory licensing” — laws that would require AI companies to pay for the journalism fed into training and generation.
Place the two stories next to each other and the message writes itself: making AI companies pay for news is leaving the territory of lawsuits and moral appeals and entering that of actual legislative drafting. For teams that train or ground models on web content, this is a variable to pencil into the cost structure now, not a debate to watch from afar.
How the Vote Went and What the Report Says
The report binds no one on its own, but it is Parliament’s official stance on generative AI and copyright. The essentials: commercial AI developers are presumed to have used copyrighted works, and the presumption can be rebutted — a softening of the committee draft, where it could not be contested; press publishers’ and broadcasters’ rights are widened so rightsholders hold “full control” over training uses of their material, and “explicit consent” is required; the text-and-data-mining (TDM) exceptions are confirmed to cover AI training, in line with the EU’s current framework and the AI Act; and the report looks into ways of dealing with infringing outputs without cutting off lawful content.
The objections came at once. Communia, a digital rights group, judged the report incoherent — it offers “no clear path forward” — and cautioned that its language on media content is sweeping enough to read press content out of the TDM exceptions, unsettling the balance EU copyright law presently strikes.
Statutory Licensing, Explained
Schiffrin and Carlini’s proposal goes a step beyond. In the working paper “How to Update EU and US Copyright Regimes in the Age of AI,” they argue for statutes that make AI companies pay publishers for journalism already consumed in training as well as for uses still to come. Pricing, in their design, should track the cost of producing news, the value of the brand, and the many purposes one article serves downstream inside AI systems — not whatever split individual platforms feel like offering.
Europe, in the two authors’ judgment, is likeliest to legislate first: the EU issued two relevant studies in 2025, and Parliament has now voted. Other designs mentioned include funds administered by the state and shared among creators, plus opt-out registries.
Where Each Country Stands
What Poynter assembles is a world pulling in several directions at once:
- Brazil: a draft bill carrying publisher payments is set for a vote in April
- Indonesia: under pressure from a trade deal with the United States, it has promised no digital taxes and no bargaining code
- South Africa: walked away from its bargaining code, settling for in-kind consideration such as Meta ad credits
- Australia: a digital levy is on the table, after Google and Meta drained the earlier News Media Bargaining Code of force
- Canada: the backing behind a proposed bargaining code is thinning out
The court dockets fill up alongside. UK publishers have banded together as the SPUR coalition to chase fair compensation; the Danish publishers’ association has taken OpenAI to court; EU publishers lodged an antitrust complaint against Google in February 2026 over AI news summaries; and in November a Spanish court levied a 524-million-euro fine on Meta for unfair competition. Danielle Coffey, who runs the News Media Alliance — roughly 2,000 news organizations strong — says it plainly: “If we get the right verdicts, we will have a functional marketplace.”
What’s Driving the Timing
Two pressures account for the moment. Evidence keeps stacking up: AI search and summary products digest news straight on the answer page, and publishers can watch their traffic and subscriptions wash away, while traditional bargaining mechanisms are losing grip quickly. Then there is path dependency: one-off licensing arrangements — Microsoft and Cloudflare have each said they will share AI revenue with publishers — reach only the large players, and statutory licensing remains the sole mechanism that sweeps the long tail of news organizations in.
AI companies get something out of it too: a cost that can finally be priced. A transparent licensing rate is easier to run than litigation risk scattered across jurisdictions — and supplying exactly that certainty is what EU lawmakers are after.
What This Means for Product Teams
Three effects land in practice. First, provenance records for training data stop being an academic courtesy: with a rebuttable presumption of use, the burden of proof shifts to the AI developer, and data pipelines must be able to answer “which content, obtained when, under which license.” Second, retrieval and summarization products stand closest to the fire — the EU publishers’ antitrust complaint over Google’s AI summaries is an early warning of the legal exposure this product shape carries, and attribution plus licensing built into the product moves from selling point to compliance floor. Third, products spanning several markets need region-by-region planning: Brazil, the EU, Australia, and Indonesia are not heading to the same answer, so content cost models will differ by region — one more input for pricing and for ranking markets.