On 11 August 2026, the Alliance de la presse d'information générale (Apig), which represents more than 300 political and general news titles, referred the unilateral rollout of AI Overviews and AI Mode to the French Competition Authority, publishers having been “presented with a fait accompli” in its words (Le Figaro), three weeks after their French launch. The first French showdown over AI answers has begun, and it tells any company monitoring its visibility one simple thing: AI summaries now capture enough value to justify litigation.
The facts
- The Apig contests the method rather than the innovation: a launch that came before negotiations were concluded, on a mere proposal to update the existing licence agreement, with a technical opt-out, and the loss of visibility that goes with it, as the only alternative. It invokes the 2019 law on neighbouring rights, which makes the use of press content conditional on prior authorisation and remuneration, and the commitments of good faith, transparency and non-discrimination made binding by the Authority since 2022 and valid until July 2027 (Le Figaro).
- The traffic losses cited by publishers: 33% to 38% on European markets according to an assessment by Arcom, and up to 42% in the United States according to the Define Media Group study of 64 sites (Blog du Modérateur).
- The precedent carries weight: in March 2024, the Authority fined Google €250 million for failing to honour these same commitments, in particular over the use of press content by its AI (Autorité de la concurrence), and its president Benoît Cœuré reiterated on 8 July that protected content used by AI Overviews must be paid for under neighbouring rights.
Why does this dispute go beyond the press?
The press is attacking first because it measures better than anyone what an AI summary costs: its traffic is its raw material, and the figures cited, a third to 42% fewer visits depending on the market, give the order of magnitude of what is at stake for any site whose audience comes from Google. France was in fact one of the last countries without AI Overviews, which had been extended to 200 countries by the end of May: the delay came precisely from the negotiations on neighbouring rights, opened after the March 2024 fine. And the Apig's central grievance, a technical opt-out as the only alternative to the fait accompli, is exactly the dilemma posed by the off switch imposed on Google by the UK regulator: everywhere, the framework for generative answers is now being negotiated before competition authorities, on top of the settings in Search Console.
The Apig itself states that “this referral in no way closes the door to negotiation”: it aims to restore the conditions for it, and points out in passing that the launch of these services confirms that press content is indispensable to generative AI. For companies outside the press, the issue is the arbitration to come rather than neighbouring rights: depending on the outcome, Google could have to adjust in France how sources are displayed, how transparent the data is or how content is paid for, all parameters that determine who gets cited and how. In the meantime, the procedure suspends nothing: AI Overviews compose their answers every day, and the only variable a company controls today is the quality of what generative engines find when they look for it.
AIxH's view
This conflict confirms, through litigation, what we measure every week: value has moved to the generated answer. Waiting for the outcome of the procedure would be the worst strategy, because the court calendar has nothing to do with the calendar of your visibility. The useful approach comes in two steps: knowing what AI Overviews, ChatGPT or Perplexity already show about your brand, then working to be cited correctly there, which our GEO agency in Luxembourg does with a panel of queries replayed at a fixed cadence. The SEO foundation remains the prerequisite: your pages are what answer engines read.