- From: Laurent Le Meur <laurent@edrlab.org>
- Date: Thu, 3 Aug 2023 12:15:40 +0200
- To: Edward Hasbrouck <ehasbrouck@nwu.org>
- Cc: "public-tdmrep@w3.org" <public-tdmrep@w3.org>
Dear Edward, There are interesting aspects in your previous comments, but I think that you draw the picture in black and white. I'm not a huge fan of the CDSM Article 4 (it was added late in the process, with insufficient considerations about implementation aspects), and not a fan of the reaction of TDM/AI actors (which are too quiet to be really honest). But we play with the cards we have in our hands and we, as technically savvy people, must try to achieve as much as possible this "balance between the interest of rightsholders and the interests of AI practitioners". I'll just respond here to your remarks below: > (1) What works have actually been copied/ingested for generative AI > development, by whom, and from what locations? > AI developers and providers of generative AI services could easily provide > a link in each derivative work generated by such a service to an online > searchable index of the works in the corpus from which it was generated, > and require in their terms of service that users include such a link. Are you really sure this is true, technically speaking? Sure that a system that ingests millions of resources can list the exact set of resources used to generate a specific new resource? I didn't study AI transformers, but I worked on categorization and recommendation systems, and I tend to have doubts about this assertion. I'd like to see proof of that. > (2) The geographic applicability of national laws and the Berne Convention: The geographic scope of the CDSM is something I asked the lawyers participating in this work. What I wrote here is my interpretation of their response. This forum is open to their remarks. What is sure is that I'd like to see something written by the EU Commission itself about the geographical scope of their work. Best regards Laurent Le Meur > Le 2 août 2023 à 17:55, Edward Hasbrouck <ehasbrouck@nwu.org> a écrit : > > Laurent Le Meur raises 2 questions: > > (1) What works have actually been copied/ingested for generative AI > development, by whom, and from what locations? > > Because generative AI developers have failed to respect the moral right of > authors to attribution, authors can only partially determine, and only by > reverse engineering, what has been scraped from the Web, by whom, and from > where. > > AI developers and providers of generative AI services could easily provide > a link in each derivative work generated by such a service to an online > searchable index of the works in the corpus from which it was generated, > and require in their terms of service that users include such a link. > > This would be similar to the requirement to include credit/attribution and > a link to the relevant Creative Commons or other open source license > whenever a derivative work includes any material under such a license. > > But AI services have not done this, probably because they know that > identifying the works in the corpus they have copied would make it easier > for the authors of works in that corpus to hold them liable for copyright > infringement or other violations of authors' rights. > > (2) The geographic applicability of national laws and the Berne Convention: > >> The second aspect has to do with the geographical limits of a European >> Directive. If a content publisher is not European and its content is >> not stored on an EU server, it seems difficult for a TDM/AI actor (EU >> or US based) to claim that such content can be scrapped under the EU >> law. And from the list of companies having signed the statement, most >> or all seem to be American organizations. I mean, it seems the claims >> you have gathered make no legal sense. And therefore the European >> Union cannot do anything about such foolish claims. I'm not a lawyer, >> this is just my understanding of the situation. > > I'm not a lawyer either, but I think this reflects a misunderstanding -- > in fact, a reversal -- of the geographic applicability of national laws > and the Berne Convention. > > In some cases national governments assert "long-arm" jurisdiction over > acts occurring outside their borders, but this is a rare exception. > > In general, jurisdiction over copyright infringement (like most other > violations of law) depends on the place where the violation occurs. > > The US has, in general, no jurisdiction over acts occurring outside the > US, even if they implicate the rights or interests of US citizens. > Similarly, the EU has, in general, no jurisdiction over acts occurring in > the US or elsewhere, even if they implicate the rights of EU citizens. > > Acts such as Web crawling carried out from the EU are governed by the laws > of the EU and its members states. If individuals outside the EU believe > that their rights have been violated by such actions, their only recourse > is under the law of the EU and/or its member states. > > If entities in the US scrape the World Wide Web, their actions are subject > to US law. If entities in an EU member state scrape the World Wide Web, > their actions are subject to the laws of that member state. > > This is why the Berne Convention was adopted: Without a treaty, citizens > of country X might have no recourse against infringing acts carried out in > country Y, if country Y chose not to protect foreign authors or works. > > The Berne Convention does not apply to national laws applicable to works > created and published within that country. A party to the Berne Convention > can choose to enact lesser protection, or none, to national works. (Some > countries including the US do, in fact, provide less protection to > national works than to foreign works.) The Berne Convention exists > specifically to set minimum standards for the protection that state > parties must provide for acts occurring in their countries that implicate > works and the rights of authors in *other* parties to the treaty. > > Activities such as cross-border Web scraping are exactly at the core of > the activities to which the Berne Convention was intended to apply. > >> Now, if your statement is about the EU DSM Directive contradicting the >> Berne Convention for content stored in Europe and/or EU content >> publishers, this is another matter. > > The Berne Convention does not apply to acts in an EU member state that > implicate works created and published in that member state. The Berne > Convention applies *only* to acts in a Berne state party that implicate > works created and/or first published in a *different* Berne state party. > > Hope this helps, > > Edward Hasbrouck > > ---------------- > Edward Hasbrouck > <ehasbrouck@nwu.org> > +1-415-824-0214 (San Francisco) > > National Writers Union > https://nwu.org > +1-212-254-0279 (New York) > > >
Received on Thursday, 3 August 2023 10:15:48 UTC