Summary
- BEREC has opened consultation on how generative AI affects internet openness, with responses due by 11 November.
- Regulators are examining how AI services select information, interact with online services and influence user choice.
- The work extends internet openness beyond network access as AI systems increasingly mediate how people discover online content.
BEREC is examining whether generative AI is changing the conditions required for an open internet as assistants, search features and generated interfaces increasingly sit between users and online information.
The Body of European Regulators for Electronic Communications opened a public consultation on 7 October after adopting a draft report on generative AI and internet openness. Responses are due by 11 November, with the regulator looking beyond the performance of individual models to the way AI systems influence access to information, services and competing sources online.
BEREC defines internet openness around the ability to access and distribute information, content and services without unlawful interference or discrimination, alongside the capacity to innovate. Generative AI complicates that model because an increasing share of online activity can begin and end inside an interface that retrieves, ranks, summarises or generates information before a user reaches the underlying service or publisher.
Traditional debates about an open internet have often concentrated on the network layer, including whether connectivity providers treat traffic fairly. AI can influence choice further up the technology stack without blocking the underlying connection at all, and BEREC notes that greater convenience and lower search costs arrive alongside limited transparency over how information is selected and prioritised.
Gateways move above the network
As generative AI is incorporated into search engines, messaging services, operating systems and other established products, large platform operators gain another mechanism through which users encounter competing information and services. A website can remain technically accessible while an AI generated answer substantially reduces the chance that a person visits it or encounters an alternative source.
BEREC is consequently examining the structure of the AI ecosystem alongside the technical protocols governing interactions between AI systems and online services. Those interfaces carry commercial consequences because publishers, software providers and other online businesses increasingly have to consider how automated systems discover, retrieve and reuse their information as well as how people find it directly.
The regulator also brings the sustainability of online content production into the analysis. Generative AI systems depend on information produced elsewhere, while generated responses can change the traffic patterns and commercial incentives that help fund the original material. Different publishers and business models will experience that shift differently, but the relationship between content supply and AI distribution is now part of the same regulatory question.
Rather than assuming a new body of AI-specific telecom regulation is necessary, BEREC is examining how existing European rules apply to the effects it has identified. That places the consultation alongside wider European scrutiny of digital platform power, competition and user choice while retaining the regulator’s focus on communications markets.
Internet access no longer ends at connectivity
BEREC’s remit originates in electronic communications, yet the practical meaning of internet access increasingly depends on software layers beyond the telecommunications network. National regulatory authorities are already using AI for tasks such as market surveillance, competition analysis, spectrum management and regulatory knowledge sharing while simultaneously having to understand how the technology alters the markets they oversee.
Generative AI services also differ substantially in architecture. Some retrieve current information from the web, others depend more heavily on knowledge encoded during training, and increasingly complex agent systems can invoke external services or tools. Source attribution, ranking, transparency and the extent of user control therefore vary considerably between products that may look superficially similar.
Those differences make a direct extension of conventional net neutrality concepts difficult. Network operators can be assessed through measurable treatment of traffic, whereas an AI system may alter outcomes through retrieval rules, ranking, model behaviour and interface design before the network connection itself becomes relevant.
Businesses that depend on online discovery will be watching whether future policy begins to address how AI gateways expose competing services and represent source material. AI developers face the corresponding question of whether users retain meaningful choice when generated interfaces are integrated deeply into operating systems, search and communications products.
BEREC has not reached a final regulatory position, and the report remains in consultation. The draft nevertheless marks a widening of the internet openness debate: access to a connection no longer determines on its own whether users can navigate a competitive online environment when software increasingly decides which information and services appear first.
