In a never-ending saga between Google and the French competition authority over copyright protection for news clippings, the Autorité de la Concurrence was announced 250 million euro fine against the tech giant on Wednesday (about $270 million at today’s exchange rate).
According to the competition watchdog, Google ignored some of its previous commitments to news publishers. However, the decision is particularly notable because it rejects something else that is up-to-date – barring Google’s use of news publishers’ content to train its Bard/Gemini AI output model.
The competition authority found fault with Google for not notifying news publishers of this GenAI use of their copyrighted content. This is in light of previous commitments made by Google to ensure fair payment conversations are held with publishers regarding the reuse of their content.
Copyright and Competition Mistakes
In 2019, the European Union approved a pan-European digital copyright reform that extended copyright protection to news headlines and excerpts. News aggregators such as Google News, Discover and the Top Stories feature box on search results pages have previously clipped and displayed these stories in their products without any financial compensation.
Google initially tried to avoid the law by disabling Google News in France. But the competition authority quickly intervened — seeing its unilateral action as an abuse of a dominant market position that risked harming publishers. The intervention effectively forced Google to cut deals with local publishers on content reuse. However, in 2021, Google was fined $592 million after the competition authority found significant violations in its negotiations with local publishers and agencies.
The tech giant called the penalty “disproportionate” and said it would appeal. But she then tried to settle the dispute — offering a series of pledges and withdrawing her appeal. The commitments have been accepted by the French Autorité, they include the transmission of key information to issuers and trading in a fair manner.
Google has signed copyright agreements with hundreds of publishers in France — which fall under the purview of its agreement with the Autorité. Therefore, its activities in this area are very strictly controlled.
No appeal
Google agreed not to challenge the Autorité’s latest findings — in exchange for a speedy process and a monetary payment.
However, its managing director of news and publishing partnerships, Sulina Connal, struck a disappointed tone — writing in long blog post that “the fine is disproportionate to the issues raised” by the authority.
The blog post suggests that Google really wants to draw a line under the saga this time, with Connal also writing: “We’ve settled because it’s time to move on, and as our many agreements with publishers show, we want to focus on the larger goal of sustainable approaches to connect people with quality content and constructive collaboration with French publishers”.
With the creation of AI in context and the competitive race to launch tools, Google’s calculus for approaching the issue of content reuse looks different.
GenAI training in context
Today’s enforcement by France’s competition authority shows it has fine-tuned Google’s use of content from news publishers and agencies for educational purposes for its AI foundation and related AI chatbot service Bard (now called Gemini).
It found that Google used content from publishers and news agencies to train Bard, its AI production tool that was released in July 2023, “without notifying copyright holders or the Authority,” according to the Press release.
At this point, Google’s defense is twofold. In its blog post it writes that the competition authority “does not challenge the way web content is used to improve newer products such as genetic artificial intelligence, which is already referred to in Article 4 of the EUCD”. [EU Copyright Directive].
Article 4 of the Copyright Directive sets out an “exemption or limitation for text and data mining” — specifically for “reproductions and exports of legally accessible works and other subject matter for the purposes of text and data mining”.
However, in its press release, the Autorité says it has not yet determined whether the exemption applies here. (It’s worth noting that the relevant clause refers to “lawfully accessible works” — while Google is legally bound to the competition authority to notify copyright holders about uses of their protected works, and apparently failed to do so in this case. )
“When it comes to declaring whether the use of news content to train an AI service falls under neighboring rights and protections, that question has yet to be answered,” the competition authority wrote. “However, the Autorité considers that Google breached its #1 commitment by failing to notify publishers that their content had been used to train Bard.”
Google’s blog post also makes a tentative reference to the EU’s AI law — suggesting it’s relevant. However, the legislation is not yet in force pending its final approval by the European Council.
The incoming AI legislation will also say that developers must comply with the bloc’s intellectual property rules. And it introduces transparency requirements with that goal in mind — requiring them to establish a policy of respecting EU copyright law. and publish a “fairly detailed summary” of the content used to train general-purpose AI models (such as Gemini/Bard).
This incoming requirement for model makers to publish a summary of training data may in the future make it easier for news publishers whose protected content has been absorbed for GenAI training to receive fair remuneration under EU copyright law.
There is no technical exception
The Autorité also points out that Google has failed to provide, at least until September 28, 2023a technical solution that allows publishers and news agencies to opt-out of their content being used for Bard training without such a decision affecting the appearance of their content on other Google services.
“Until that date, publishers and news outlets that wanted to opt out of this use case had to enter a directive that blocks all content indexing by Google, including Google Search, Discover and News services. These services are specifically part of the negotiation for revenues related to neighboring rights,” he wrote, adding: “Going forward, the Autorité will carefully consider the effectiveness of Google’s opt-out procedures.”
In more technical terms, between July and September 2023, news publishers could insert a “noindex” tag in their robots.txt file to make sure their content wasn’t used to train Google’s AI model. This robots.txt file is placed in the root folder of web servers and contains various instructions for search engines. Google’s web crawler examines the instructions in these files to index websites.
But a “noindex” tag means your site disappears from Google entirely. In September 2023, Google added more granularity and created a “Google-Extended” rule that differs from the “noindex” rule. By opting out of the Google-Extended directive, web publishers indicate that they do not want to help improve the current and future model of Gemini.
Other deficiencies
The Autorité also fines Google for a number of other issues related to the way it negotiates with French news publishers, finding that it failed to provide them with all the information needed to ensure fair negotiation of remuneration for their content.
Inside Press releasewrote that Google’s information to publishers about its methodology for calculating the amount to be paid was “particularly opaque”.
It also found that Google had failed to meet non-discrimination criteria aimed at ensuring equal treatment of publishers. And it called Google’s decision to impose a “minimum threshold” on the fee — meaning below which it would not make payments to publishers — with the Autorité describing it as introducing discrimination between publishers “at its inception”. Below a certain threshold all publishers are “arbitrarily allocated zero remuneration, regardless of their respective status,” its press release also noted.
In addition, the Autorité found fault with Google’s calculations of so-called “indirect income”, saying that the “package” it proposed was not in line with previous decisions or the Court of Appeal decision, from October 2020.
It also said that Google did not honor its commitment to update the remuneration contracts in accordance with its commitments.