Brief IA

Google's AI Overviews: A Threat to French Journalism

🛠️ AI Tools·Tom Levy·

Google's AI Overviews: A Threat to French Journalism

Google's AI Overviews: A Threat to French Journalism
Key Takeaways
1Google will launch its AI Overviews in France on July 22, 2026.
2In the UK, this feature has reduced traffic to publisher sites.
3The French press could see its online audience affected in a similar way.
💡Why it mattersAI Overviews could transform traffic to news sites in France.
Le brief IA que lisent les pros

Le brief IA que les pros lisent chaque soir

Les 7 actus IA du jour, décryptées en 5 min. Gratuit.

Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.

Choisis ton rythme

Gratuit · Pas de spam · Désabonnement en 1 clic

📄
Full Analysis

Google AI Overviews: A Threat to the French Press

France, the Last Bastion to Fall

Type a query, browse through a dozen blue links, or maybe two, click on the one that seems most relevant, and find the answer or service you were looking for. For a quarter of a century now, Google has operated this way: a search engine that directs users to the page in its index that it deems "the most relevant and qualitative" based on their query. Could this foundational principle be shaken? And with it, the entire economic model of the press, and then the web as a whole?

Since the launch of AI Overviews in the United States in May 2024, and then in over 200 countries, this question has arisen. Activated by default and placed at the top of the results, these summaries are generated by a Gemini model adapted for Search, which scours indexed pages and provides a synthesis of a few lines, accompanied by links to the selected sources. The result: most users now get answers to their queries directly on the search results page, without needing to visit an external site. In other words, the company from Mountain View is uniquely managing to keep its users within its ecosystem.

This shift, where Google would stop acting as the switchboard of the web, was predicted by some long before the AI Overviews or ChatGPT, at a time when traffic from the search engine was reaching peaks for most publishers. Nilay Patel, editor-in-chief of the tech media The Verge, is among them. After witnessing firsthand the decline in referrals from Facebook, he predicted as early as the pandemic that Google would eventually "turn off the tap." "I was thought to be crazy," he confided to the New York Times in 2023. He even named this prophecy: Google Zero. In an article published in May 2024, he admitted he never stopped questioning media leaders on the effects this scenario would have on their business in his podcast Decoder. He even went so far as to confront Sundar Pichai, Google's CEO, on the subject. "No one provided me with a satisfactory answer, and it seems the media industry still thinks it can manage the problem when the time comes," he wrote. He, at least, has completely rethought his media's model by introducing a paywall.

The British Laboratory: A Bitter Preview

In France, the prophecy is about to take a very concrete turn. In a statement, Google announced that the AI Overviews and their conversational variant, AI Mode, are being rolled out in France starting Wednesday, July 22, confirming information revealed at the end of June by Ouest-France. This decision ends two years of uncertainty, during which French media were left to measure the potentially devastating consequences of the feature by anxiously scrutinizing studies conducted across the Atlantic, which reported sometimes staggering drops in traffic.

Since May 2024, France had been an exception in the landscape: one of the few major markets to escape these AI-generated summaries, or to be deprived of them, depending on which side of the fence one stands. The issue at hand: neighboring rights, a legal framework introduced by the law of July 24, 2019, which requires platforms to compensate publishers for the reuse of their content. Historically hostile to this principle, the Mountain View firm has been called out multiple times for evading its obligations. It was fined 500 million euros in July 2021 for failing to "negotiate in good faith with publishers," and then received an additional fine of 250 million euros in March 2024 for failing to meet some of its previous commitments, particularly regarding transparency. It was also criticized for not informing publishers about the use of their content to train AI models that, at the time, powered its conversational agent Bard.

The letter that Ouest-France accessed at the end of June formalizes a compromise, intended to remove what Google perceives as a regulatory obstacle, even as the company believes it already complies better with the rules than its competitors. "Google is today one of the only, if not the only, companies to pay neighboring rights to all types of press, for very significant amounts," claimed Sébastien Missoffe, managing director of Google France, in the columns of the regional daily.

Three commitments accompany the rollout:

  • A right to withdraw, which allows each publisher the choice to appear or not in the AI Overviews, without this refusal affecting their ranking in search results.

  • Increased transparency, with metrics separating impressions from AI-powered features from those of traditional search.

  • Compensation, finally, under neighboring rights for the 450 publishers under agreement with the American company, as long as their content displayed in these summaries is viewed by users.

Several of these obligations, including the right to withdraw, echo those imposed on Google across the Channel, in early June, by the Competition and Markets Authority (CMA), the British competition authority. Under pressure from the regulator, the company announced on the same day that it would test these features with a limited number of publishers in the United Kingdom before rolling them out globally. This partly explains the relative ease with which Google complied in France after reaching an agreement on compensation.

The United Kingdom is not only paving the way on the regulatory front: it also provides, with nearly two years of hindsight, an initial assessment of what the AI Overviews produce for publishers.

AI Overviews: On Google, Click-Through Rates for Top Positions Drop by 32%

AI Overviews appeared across the Channel in August 2024, three months after their launch in the United States. However, their actual penetration has long been difficult to assess in this market where Google commands over 90% of web searches. "It was not possible to measure the extent of their dissemination in the UK before the beginning of April this year, when Google began testing the AIO with non-logged-in users," contextualized DMG Media, owner of the Daily Mail and Metro, in a document sent to the British competition authority.

In April 2025, Authoritas, a platform specializing in AI and SEO, published one of the first comprehensive studies dedicated to their penetration and impact in the UK. Conducted between April 16 and 22 on a corpus of keywords and phrases related to topics covered by major British headlines, it noted a 47.5% drop in click-through rates on desktop and a 37.7% drop on mobile as soon as an AI-generated summary appeared at the top of the results, reports Press Gazette. Out of 3,500 queries, AI Overviews triggered for 426 terms on desktop and 427 on mobile, bringing the average penetration rate to 12.2% across both platforms.

It is already serious enough that Google is content to steal journalists' work and pass it off as its own.

Brief IA — L'actualité IA en français

L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.