France’s competition authority has put Google’s relationship with news publishers back under scrutiny, this time with generative AI at the center of the dispute. The Autorité de la Concurrence announced a €250 million fine after finding that Google failed to honor commitments linked to negotiations over publisher content.
The case matters beyond France because it connects two issues that are increasingly hard to separate: how platforms pay for news content, and how AI companies use that content to train new products. In this decision, the authority focused not only on Google’s bargaining conduct, but also on the use of publisher and press agency material for Bard, now called Gemini.
Why France fined Google again
The conflict sits inside a longer fight over copyright protections for news headlines and snippets. In 2019, the European Union passed a pan-EU digital copyright reform that extended protections to these forms of news content.
Before that reform, services such as Google News, Discover and the “Top Stories” feature box on search results pages had displayed news stories without financial compensation to publishers. Google initially tried to avoid the French application of the rule by switching off Google News in France.
The Autorité de la Concurrence intervened, finding that Google’s unilateral move was an abuse of a dominant market position that risked harm to publishers. That intervention pushed Google toward deals with local publishers over content reuse.
The dispute did not end there. In 2021, Google was hit with a $592 million fine after the authority found major breaches in negotiations with local publishers and agencies. Google called that sanction “disproportionate” and said it would appeal, but later sought to settle, offered a series of pledges and withdrew its appeal.
Those accepted commitments included passing key information to publishers and negotiating in a fair way. Google has since signed copyright agreements with hundreds of publishers in France, placing this part of its business under close regulatory supervision.
How Bard and Gemini changed the stakes
The latest decision is notable because it brings AI training directly into a copyright and competition enforcement dispute. The authority found that Google used content from publishers and press agencies for training Bard, its generative AI tool launched in July 2023, without notifying copyright holders or the authority.
Google’s position, according to its blog post, is that the authority “does not challenge the way web content is used to improve newer products like generative AI, which is already addressed in Article 4 of the EUCD”. Article 4 of the Copyright Directive concerns an exception or limitation for text and data mining involving lawfully accessible works.
The Autorité de la Concurrence did not accept that the issue was settled. It said the question of whether using news content to train an artificial intelligence service falls under neighboring rights and protection has not yet been answered.
That distinction is important. The authority’s finding was not framed only as a broad judgment on AI training. It was tied to Google’s existing commitment to inform publishers about uses of their protected works.
In other words, the AI issue landed inside an already regulated relationship. Google was not dealing with publisher content in a blank legal space; it was operating under commitments that the French authority had already accepted and was monitoring.
The opt-out problem for publishers
The authority also criticized Google’s opt-out process. Until at least September 28, 2023, it said Google did not provide a technical solution that allowed publishers and press agencies to prevent their content from being used to train Bard without also affecting display on other Google services.
That created a difficult choice for publishers. According to the authority, those that wanted to opt out had to use an instruction that blocked all content indexation from Google, including Search, Discover and Google News.
In technical terms, between July and September 2023, publishers could use a “noindex” tag in the robots.txt file to keep content from being used for Google’s AI model. But that also meant disappearing from Google altogether.
In September 2023, Google introduced more granular control through a “Google-Extended” rule, separate from the “noindex” rule. By opting out through Google-Extended, web publishers could indicate that they did not want to help improve Gemini’s current and future models.
The Autorité said it will carefully review the effectiveness of Google’s opt-out processes in the future. That point shows how much the practical mechanics of AI consent now matter: a right to object is far less useful if exercising it also removes a publisher from core discovery services.
Payment talks remain under pressure
The AI training issue was not the only finding. The authority also sanctioned Google over other shortcomings in how it negotiated with French news publishers.
It found that Google failed to provide all the information needed for fair bargaining over remuneration. In its press release, the authority described Google’s information about the methodology for calculating payment as “particularly opaque”.
The authority also found problems with non-discrimination criteria, which were intended to ensure equal treatment for publishers. It objected to Google imposing a “minimum threshold” for remuneration, below which publishers would receive no payment.
According to the authority, that approach introduced discrimination between publishers “in its very principle”. It said that below a certain threshold, publishers were allocated zero remuneration regardless of their individual situation.
The authority also challenged Google’s calculations regarding “indirect income”, saying the package proposed was not in line with previous decisions or the appeal judgment of the Court of Justice from October 2020. It also said Google failed to update remuneration contracts in line with its commitments.
Why the decision matters
Google agreed not to contest the latest findings in exchange for a fast-tracked process and a monetary payment. Sulina Connal, Google’s managing director for news and publishing partnerships, wrote that “the fine is not proportionate to issues raised” by the authority.
She also wrote: “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 connecting people with quality content and on working constructively with French publishers.”
The settlement may close this enforcement round, but the underlying questions remain active. Generative AI has made publisher content more valuable and more contested, while existing copyright commitments still shape what large platforms can do with that content.
The decision also points toward coming transparency pressure. The source article notes that the EU AI Act is pending final adoption by the European Council and will require developers to follow the bloc’s copyright rules. It will also introduce transparency requirements, including a policy to respect EU copyright law and a publicly available, sufficiently detailed summary of content used to train general purpose AI models such as Gemini/Bard.
For publishers, the French case shows that AI training is no longer a separate technical matter. It is becoming part of the same bargaining, disclosure and payment debate that has shaped the relationship between news organizations and dominant platforms for years.