Publishers facing AI Overviews: in the United Kingdom, the CMA creates a technical opt-out, not a right to payment — ActuIA

April 4, 2022 | Source: ActuIA
Create a watch on this topic Receive upcoming developments if any, with an AI watch configured from this article.
Create my free watch

Between a contractual licensing regime inherited from neighboring rights and an AI-specific opt-out mechanism, the choice of public policy is not the same - and it is precisely this that the CMA order of June 3, 2026, makes visible. Until now, a British publisher who wanted to prevent Google from siphoning off its articles to feed the AI Overviews had only one lever: completely withdrawing from natural referencing - a prohibitive choice given the engine's dominant position. The formal order issued on that day to Alphabet by the Competition and Markets Authority (CMA), under the Digital Markets, Competition and Consumers Act 2024, separates the two: publishers will be able to block AI usage without losing their visibility in search, according to the CMA order of June 3, 2026. Google had, in March 2026, indicated it was exploring such a mechanism - the novelty lies in its opposability, not its principle.

An ex ante instrument, not a retrospective sanction

The Digital Markets, Competition and Consumers Act 2024 empowers the CMA to designate companies with Strategic Market Status (SMS) and impose binding Conduct Requirements on them. The sequence applied to Google is consistent with this pattern - public consultation on two proposed obligations (a Publisher Conduct Requirement regarding opt-out from AI Overviews and a Fair Ranking Conduct Requirement) was opened on January 28, 2026, followed by a formal order on June 3, 2026, which is the first Conduct Requirement adopted under the DMCC Act 2024 regime on search services according to Bloomberg. The precise technical details - triggering the opt-out, scope of "AI features" covered, compliance timeline, penalties - are not detailed in the provided material and remain to be confirmed in the official publication by the authority.

What the UK precedent shifts in the EU toolkit

In the Union, two trajectories coexist without merging. The first is procedural: according to the preliminary conclusions of the Commission on April 16, 2026, Brussels addressed Google, under the Digital Markets Act, with measures concerning the sharing of search data with third-party engines and, notably, with AI chatbots equipped with search functionalities. The subject differs from that of the CMA - removing data rather than isolating the use of AI in indexing - but the thematic convergence is clear. The second trajectory is legislative: the 2019/790 directive, known as DSM, in its Article 15, establishes a right similar to that of press publishers, whose interaction with AI Overviews has not yet been clarified at this stage. The UK order does not fall under either of these two frameworks and is not legally binding within the EU order: it serves as a comparative reference, not as a source of law. It demonstrates, however, the technical feasibility of a sector-specific opt-out decoupled from de-indexing - an element that the Commission may observe in the design of DMA remedies, without being compelled to do so.

«World first»: what the formula covers, and what it overlooks

The CMA refers to its own action as «world first». Strictly read - as the first binding Conduct Requirement specifically targeting the opt-out of integrated generative summaries within a dominant search engine - the formula is defensible. Broadly read, however, it erases a precedent well known to European editors: the French transposition of the DSM directive forced Google to negotiate compensation for online press as early as 2019, and a comparable mechanism was implemented in Germany. The scope, however, was not that of AI Overviews - it concerned the display of snippets and editorial republication within Google News services, not the feeding of a generative model integrated into the search engine. The distinction is not trivial for publishers: neighbouring rights led to licensing agreements; the British Publisher Conduct Requirement, on the other hand, produces a technical blocking mechanism upfront, which does not preclude compensation. The nuance illuminates the upcoming choices, on the continent as well as across the Channel: between a contractual regime inherited from online press and an opt-out specifically applicable to AI use, the choice of public policy is not the same.

Add ActuIA to my preferred sources on Google

Our articles will appear first in Google's «On the Front Page».

Also read
  • Nouvelle étape pour l'AI Act : les obligations pour les modèles d'IA à usage général sont entrées en vigueur Law / adopted text New step for the AI Act: obligations for general-purpose AI models are now in effect August 4, 2025
  • Quand consulter le CSE avant de déployer l'IA (octobre 2026) File / background analysis When to consult the CSE before deploying AI (October 2026) October 7, 2026
  • Décider de l'IA à l'école sans attendre l'essai randomisé, l'avis anglais Forum / opinion Deciding on AI in schools without waiting for the randomized trial, the English opinion October 2, 2026
  • AI Act : les enquêtes pourraient prolonger le délai de sanction Bill / consultation AI Act: investigations could extend the sanction period October 15, 2026
In the watch
  • November 4 According to the Washington Post, Google DeepMind employees are challenging lifetime non-disclosure clauses and want to be able to alert about the risks of AI
  • November 4 According to AP and the New York Times, schools that refuse Google's AI tools will also lose access to Google research on student accounts
  • November 4 CJUE: for Advocate General Szpunar, the responses of a chatbot that reuses press articles may violate the rights of publishers, not the model's training
All the watch →
ST
Stephane Nachez

Editing ActuIA - news, data and analyses on artificial intelligence for decision-makers.