Les générateurs d’images IA actuels comme Midjourney, Dall-E et j’en passe, font polémique puisqu’ils piochent allégrement dans les œuvres des artistes sans leur consentement ni rémunération. Mais des solutions sont en train de se mettre en place pour les entreprises qui souhaiteraient utiliser l’IA pour illustrer leurs supports sans pour autant piller les artistes.

Une de ces solutions, c’est Tess qui propose une approche que j’ai trouvée intéressante puisqu’elle met en place des collaborations avec des créateurs pour utiliser leur style dans des modèles de diffusion d’images.

Concrètement, chaque modèle d’IA est entraîné sur le style visuel d’un artiste unique, avec son accord puis l’outil utilise un SDXL custom (Stable Diffusion) pour générer les images. Évidemment, à chaque fois qu’une image est générée avec son style, l’artiste touche des royalties ! C’est gagnant-gagnant donc.

L’outil intègre également un système de métadonnées basé sur le protocole C2PA , qui permet d’identifier clairement les images générées par IA et de les distinguer des créations originales.

L’objectif de Tess est donc clair : démocratiser la création d’images artistiques de qualité, en la rendant accessible au plus grand nombre, tout en rémunérant équitablement les artistes et leur permettant de garder la maitrise de leur art. C’est une bonne idée vous ne trouvez pas ?

Et les artistes dans tout ça ? Et bien si j’en crois le site de Tess, ils sont déjà plus de 100 à avoir déjà sauté le pas, séduits par ce nouveau modèle de rémunération. Maintenant si vous voulez devenir clients de la plateforme, ce n’est malheureusement pas encore ouvert à tous, mais une liste d’attente est en place pour vous inscrire et être tenu au courant.

Ça va en faire des déçus chez les petits malins qui croyaient avoir trouvé la combine du siècle pour se faire du blé facile sur Medium en balançant des articles pondus par ChatGPT et compagnie ! La plateforme de blogs vient en effet de dégainer son bazooka anti-IA et de bannir purement et simplement le contenu généré artificiellement de son programme partenaire rémunéré. Eh oui, fini de laisser tourner GPT-4 à plein régime toute la nuit pour cracher des articles à la chaîne et les planquer derrière un paywall ! Medium a dit « Niet, c’est fini ça ! «

La sentence est tombée par mail dans la boîte des utilisateurs : à partir du 1er mai , c’est tolérance zéro pour les histoires 100% IA dans le programme de monétisation. Les récits conçus par une intelligence artificielle seront donc retirés des paywalls illico presto, et les comptes récidivistes pourront même se faire virer de la rémunération par Medium. Ça rigole plus !

Faut dire que la plateforme a une vision bien précise de son identité : un sanctuaire pour le storytelling humain , pas un repaire de scribouillards synthétiques.

Ok, ils sont ouverts à ce que l’IA file un coup de main pour peaufiner un texte ou aider à écrire dans la langue de Shakespeare quand on est rouillé en anglais. Mais que dalle pour les papiers écrits de A à Z par un algorithme, c’est no way !

« Medium, c’est fait pour les histoires humaines, pas pour les textes générés par l’IA « , martèle la plateforme dans son mail qui a dû faire l’effet d’une douche froide à plus d’un. En fait, si on y regardait de plus près, c’était déjà écrit noir sur blanc dans les guidelines de Medium : les histoires 100% IA, c’était déjà restreint à une diffusion limitée au petit cercle de l’auteur sur la plateforme. Pour utiliser l’IA en mode fair-play , faut jouer la transparence et le mentionner direct dans les 2 premiers paragraphes. Pareil pour les images générées par l’IA, qui doivent être identifiées et sourcées comme il faut.

Mais là, en virant le contenu IA de son programme de monétisation, Medium serre sérieusement la vis. Et si jamais vous tombez sur un de ces récits bricolés par un bot et que ça vous gonfle, pas de souci ! Il suffit de cliquer sur « Voir moins de contenu similaire » pour dire à Medium « Merci mais non merci, très peu pour moi ! » et ainsi limiter la propagation de ces histoires synthétiques.

Après soyons honnêtes, le contenu généré par l’IA, c’est souvent mal fait et on se retrouve avec une belle bouillie insipide qui manque de saveur. Étant abonné payant à Medium, je trouve que c’est pas une mauvaise nouvelle et j’imagine que bon nombre de plateformes vont leur emboiter le pas.

Source

Le géant américain n'abandonne pas son idée de monde virtuel, mais a émis de sérieux doutes quant aux intelligence artificielles génératives actuellement présentes sur le marché.

The Internet Archive (IA) is a non-profit organization that aims to preserve digital history for generations to come.

The digital library is a staunch supporter of a free and open Internet and began meticulously archiving the web over a quarter century ago.

In addition to archiving the web, IA also operates a library that offers a broad collection of digital media, including books. Staying true to the centuries-old library concept, IA patrons can also borrow books that are scanned and digitized in-house.

Publishers vs. Internet Archive

The self-scanning service offered by the Internet Archive (IA) differs from the licensing agreements entered into by other libraries. Not all publishers are happy with IA’s approach, resulting in a major legal battle two years ago.

Publishers Hachette, HarperCollins, John Wiley, and Penguin Random House filed a lawsuit, equating IA’s controlled digital lending (CDL) operation to copyright infringement. Earlier this year a New York federal court concluded that the library is indeed liable for copyright infringement .

The court’s decision effectively put an end to IA’s self-scanning library, at least for books from the publishers in suit. However, IA is not letting this go without a fight and in December the non-profit filed its opening brief at the Second Circuit Court of Appeals, hoping to reverse the judgment .

High Profile Support

The importance of this legal battle is illustrated by the large number of amicus briefs that are filed by third parties. Previously, IA received support from copyright scholars and the Authors Alliance, among others.

A few days ago, another round of amicus came in at the Court of Appeals, this time to back the publishers who filed their reply last week . In more than a handful of filings, prominent individuals and organizations urge the Appeals Court not to reverse the district court ruling, arguing that this would severely hamper the interests of copyright holders.

The briefs include positions from industry groups such as the MPA, RIAA, IFPI, Copyright Alliance, the Authors Guild, various writers unions, and many others. Legal scholars, professors, and former government officials, also chimed in.

RIAA, MPA, et al.

The RIAA and MPA submitted an amicus brief together with the NMPA and the News Media Alliance. These industry groups draw a parallel between the impact Napster and BitTorrent had on music and movie sales, and the threat IA’s self-scanning library poses today.

“Digital piracy has inflicted a huge economic toll on those industries and, by extension, on their ability to invest in new creative works and the artists who make them. Internet Archive’s theory of fair use represents a threat just as grave.”

Industry groups fear that if the Internet Archive is allowed to digitize and lend books, it could set a precedent for other forms of media. For instance, if services were able to lend music, movies, or news media to the general public, these industries might face similar challenges.

“Deeming Internet Archive’s mass reproduction and distribution program to be fair use would no doubt embolden not only Internet Archive itself but also other online platforms to freely ‘lend’ all types of copyrighted works to the public in digital formats,” they write.

“That would catastrophically harm the digital markets on which the music industry, the movie and television industry, the news industry, and similar industries depend to profitably create and distribute their works—and would thereby undermine the incentive for the creation of new works that copyright law exists to protect.”

According to the amici, there is nothing fair about IA’s digital library; instead, they see it as “unambiguous copyright infringement.”

Copyright Experts, Professors, and Lawmakers

A second amicus brief is submitted by more than a dozen professors and scholars of copyright and intellectual property law. They stress that IA’s practice should not be seen as “transformative” fair use, arguing that the library offers a “substitution” for books that are legally offered by the publishers.

This sets the case apart from current legal precedents including the Google Books case , where Google’s mass use of copyrighted books was deemed fair use.

“IA’s exploitation of copyrighted books is thus the polar opposite of the copying that was found to be transformative in Google Books and HathiTrust . IA offers no ‘utility-expanding’ searchable database to its subscribers. What it does offer is access to full-text books as a clearly competing substitute for the versions licensed by book publishers,” the legal scholars write.

Another amicus brief adds more heavyweight support for the publishers. This includes former judges and two dozen government officials and lawmakers, including Lamar Smith, former Chair of the House Judiciary Committee, and Bob Goodlatte, former Chair of the House Judiciary Committee.

This brief also rejects the Internet Archive’s fair use arguments, framing the library as a threat instead.

“IA does not further the public interest, but rather undermines incentives to create and disseminate books that benefit society. Thus, its actions are decidedly not protected by fair use,” their brief reads.

IA and AI

The final amicus brief we want to highlight comes from a broad collection international and regional trade groups from outside the United States. These include the International Publishers Association, the International Video Federation, and the Association of Canadian Publishers.

These groups also reject the fair use arguments. They stress that in addition to directly competing with the interests of publishers, IA’s library is also an indirect ‘artificial intelligence’ threat as the digitized books can be used as AI training material.

“The Internet Archive is an obvious source of high-quality works for AI training since these works have been professionally edited and improved by publishers. Entering the terms ‘Internet Archive DRM’ into any search engine results in a number of links to software tools that remove the Internet Archive’s DRM technology along with instructions on how to use it.

“Even if AI training is ultimately determined by U.S. courts to not be a fair use, Amici fear that the Internet Archive’s CDL collection has already been used as an AI training tool,” the international trade groups add.

In summary, the book publishers have plenty of external support for their legal battle. However, it remains to be seen whether any of these amici, including those in favor of IA, will influence the eventual outcome of the appeal.

—

Below is an overview of the amicus briefs that were filed over the past few days, all in support of the publishers.

– RIAA, MPA, NMPA, News Media Alliance .
– Former government officials, former judges, and intellectual property scholars .
– Copyright Alliance .
– Various organizations that represent the interests of writers and other creators .
– Professors and scholars of copyright and intellectual property law .
– International and regional trade bodies .

From: TF , for the latest news on copyright battles, piracy and more.

In 2020, publishers Hachette, HarperCollins, John Wiley and Penguin Random House sued the Internet Archive (IA) for copyright infringement, equating its ‘Open Library’ to a pirate site.

IA’s library operates as a non-profit organization that scans physical books, which can then be loaned out in an ebook format. Patrons can also borrow books that are scanned and digitized in-house, with technical restrictions that prevent copying.

Staying true to the centuries-old library concept, only one patron at a time can rent a digital copy of a physical book. These restrictions were temporarily loosened at the height of the COVID-19 epidemic when IA launched the National Emergency Library.

Publishers vs. Internet Archive

The self-scanning approach differs from the licensing deals other libraries enter into. Not all publishers are happy with IA’s service which triggered a massive legal battle four years ago.

Publishers Hachette, HarperCollins, John Wiley, and Penguin Random House filed a lawsuit, equating IA’s controlled digital lending (CDL) program to a piracy operation. Last year, a New York Federal court concluded that the library is indeed liable for copyright infringement .

The Court’s decision effectively put an end to IA’s self-scanning library, at least for books from the publishers in suit, but IA isn’t prepared to let go without a fight. Last December, the non-profit filed its opening brief at the Second Circuit Court of Appeals, hoping to reverse the judgment .

Publishers Respond to IA’s Appeal

IA argued that its scanning-and-lending activities amount to fair use. They cited expert witnesses who concluded that there’s no financial harm and further argued that the service is substantially different from the ebook licensing market.

Late last week, the publishers filed a redacted copy of their reply brief at the Second Circuit Court of Appeals. According to the companies, IA engaged in “massscale infringement” by making digital copies of physical books without permission from the works’ rightful owners.

The IA-digitized books are then lent to patrons instead of officially-licensed ebooks, which the publishers see as a direct threat to their rights and business.

“Controlled digital lending is a frontal assault on the foundational copyright principle that rightsholders exclusively control the terms of sale for every different format of their work – a principle that has spawned the broad diversity in formats of books, movies, television and music that consumers enjoy today,” the publishers write.

The publishers state that more than 93% of the public libraries in the US license ebooks through official channels. IA refuses to do so, arguing that making their own digital copies of legitimately purchased books amounts to fair use.

Not a VCR

IA views these digital copies as “transformative” and argues that its scan-and-digitize practice is fair use. In the appeal brief, the library equated it to the digital copies of video broadcasts, which people can legally make following the Betamax decision .

The publishers, however, maintain that IA’s lending operation is an effort to bypass official licensing channels, arguing that it has little to do with how people could use a VCR for private use.

“IA reaches far afield and twists Sony beyond recognition in an effort to manufacture some support. But Sony only held that it was fair use for users of Betamax machines to ‘time-shift’ free television programs ‘for private home use’.

“This bears no resemblance to IA’s massive book digitization project systematically distributing bootleg ebooks to the worldwide general public,” the publishers add.

The book publishers believe that IA is headed into uncharted waters, where no other library has ever gone. While IA sees its lending program as a logical and widely accepted step, the rightsholders frame it as a novel digital threat.

“Public libraries […] have never engaged in mass-digitization of millions of commercially available print books and distributed the resulting ebooks to anyone with an internet connection; they have never systematically evaded publishers’ terms of sale for specific formats like ebooks; they have never partnered with a used bookstore to funnel books to offshore scanning facilities..,” the publishers write.

All About Control

IA and several supporters , including the Authors Alliance and various copyright scholars, previously argued that publishers currently have too much power and control.

Allowing libraries to scan and digitize their books would certainly limit this control, but the publishers believe that’s clearly against the law.

“In short, IA’s practice of CDL is radical and unlawful. A decision deeming CDL fair use would have a dire impact on book publishing and all creative industries. Libraries around the country could skirt the current library ebook markets, fundamentally interfering with the Publishers’ digital strategies and destabilizing book markets,” they write.

A Napster Moment?

The publishers fear a ‘Napster moment’ for books if outsiders can run their own digitization programs and operate distribution platforms, without the involvement of rightsholders. Napster has shown that this is a viable threat and recent legal discussions surrounding the use of copyrighted works for AI training make this a very relevant issue today.

“Indeed, as technology companies ‘train’ generative AI products on vast numbers of books and other media, maintaining legal protection for derivative uses has never been more important,” the publishers note.

“The long-range disruptions in the music industry caused by Napster and other file sharers are a cautionary tale on the dangers of illicit copying that deprives rightsholders of the ability to control their markets.”

Interestingly, Napster’s existence arguably provoked the foundation of a new music industry model. It proved to be a large inspiration for innovators including Spotify’s Daniel Ek , who pioneered the music streaming business that generates most of the industry’s revenue today.

The publishers don’t elaborate on the broader implications of Napster, but they likely refer to the drastic decline in music sales that took place shortly after the service gained popularity.

The publishers ask the court to affirm the lower court’s decision, which currently prevents IA from lending out digitized copies of its books. The request is backed by a series of detailed legal arguments and citations in the full brief, which can be accessed here (pdf) .

From: TF , for the latest news on copyright battles, piracy and more.

Les élèves ne sont pas les seuls à miser sur l'IA pour éviter de faire leur travail.

Les plateformes sociales stars des années 2000 se sont trouvées une nouvelle manne financière à l'éthique discutable.

Mattel souhaite révolutionner son grand classique à coup d'intelligence artificielle mais les premiers retours sont mitigés.

Quoi de plus pratique pour converser avec un chatbot que de lui envoyer un SMS ?