Cyber Territories — Dispatch #10
The Forgotten Infrastructure: Why Democracies Cannot Afford to Lose Their News Agencies
The idea for this Dispatch emerged during a recent European Alliance of News Agencies (EANA) meeting at the European Parliament, where discussions on AI, media freedom and democratic resilience repeatedly returned to the same uncomfortable question: how did Europe allow one of its most strategic infrastructures, professional news agency production, to become economically dependent on systems designed and controlled elsewhere?
For more than twenty years, European publishers and news agencies adapted themselves to a digital economy increasingly shaped by algorithmic advertising logic. The promise appeared simple. News organisations would provide quality journalism, while technology platforms would provide audience reach, traffic and advertising revenue. In practice, the opposite happened. Advertising revenues migrated steadily from editorial organisations toward automated advertising systems operated by a small number of global platforms, while publishers and agencies were left competing over shrinking margins in markets they had originally helped create.
This structural imbalance is analysed in detail in Google v News Publishers and IP Rights by Jeroen Huijts, Copenhagen Business School, which examines the asymmetry between platform economics and publisher dependency in the digital ecosystem.
News agencies occupied a paradoxical position inside that transformation. They continued investing in verified reporting, agenda services, photo streams, data blocks and international coverage while simultaneously watching their main clients, publishers and broadcasters, lose the economic capacity to sustain the very infrastructure upon which modern journalism depends.
The result is a structural contradiction. Societies increasingly rely on trusted information for political, economic and security decision-making, while the organisations producing verified information operate inside economic models optimised primarily for advertising extraction and engagement metrics.
From Search Engine to Final Destination
In April 2026, the European Parliament’s EUDS Special Committee published a report that framed the emergence of Google AI Overviews not as a cosmetic redesign of search interfaces, but as a structural shift in how citizens access information and how informational value is distributed online.
The report documents how search engines are evolving from intermediaries into destinations. Instead of directing users toward original sources, AI-generated summaries increasingly provide synthetic answers directly inside platform environments. Publishers remain the underlying source layer, but users no longer need to visit them.
The economic consequences are already measurable. Studies cited in the report indicate that AI-generated overview layers can reduce click-through rates to news publishers dramatically. Traditional first-position search results are pushed downward, while “zero-click searches” continue to rise. The user receives an answer, but the original editorial ecosystem loses visibility, traffic and ultimately revenue.
This evolution accelerates an already fragile economic reality.
Economic Erosion Before AI
Long before generative AI appeared, the digital advertising market had already become structurally distorted. Large technology platforms gradually absorbed the overwhelming majority of digital advertising growth while publishers lost classified advertising, audience ownership and pricing power.
The consequence was not merely commercial pressure. It was infrastructural weakening.
Professional journalism is expensive. International reporting, verification, photography, fact-checking and legal oversight require permanent investment. News agencies absorb part of these costs centrally for entire national ecosystems. When publishers lose economic capacity, cuts eventually reach the agencies themselves.
The conflict between publishers and platforms predates generative AI by many years. The Belgian Copiepresse litigation against Google already revealed the structural imbalance between search visibility and publisher dependency.
Copiepresse vs Google: legal proceedings and publisher rights
Media freedom organisations continue to warn about excessive concentration inside digital advertising markets.
The central issue is no longer difficult to identify. One side controls the advertising infrastructure, auction systems, data collection and algorithmic distribution. The other side produces the content while competing for declining residual revenues.
AI Overviews as an Accelerator
Generative AI did not create this crisis. It accelerated it.
The European Parliament report demonstrates that AI Overviews intensify existing traffic erosion by transforming professional journalism into a background resource layer feeding synthetic interfaces.
Press Gazette recently analysed the likely impact on publisher visibility.
Devastating potential impact of Google AI Overviews on publisher visibility revealed — Press Gazette
Smaller publishers, regional media and organisations operating in medium-sized language markets face disproportionate risks. They lack the brand power and direct audience relationships required to survive large-scale disintermediation. In multilingual Europe, this directly impacts media pluralism.
The issue extends beyond economics. Visibility itself becomes concentrated.
If citizens increasingly consume synthetic summaries generated by a limited number of AI systems, informational diversity risks being compressed into a smaller number of algorithmically prioritised narratives.
The Regulatory Blind Spot
European regulation has not fully adapted to this transition.
The AI Act focuses primarily on training obligations and high-risk systems. The DSM Directive addresses copyright and neighbouring rights. The European Media Freedom Act focuses largely on ownership concentration and political interference.
None of these frameworks adequately address AI-mediated informational disintermediation during inference and answer generation.
This gap matters enormously.
The practical problem is no longer limited to whether models were trained on journalistic content. The issue is how professional journalism is continuously reused, synthesised and monetised inside answer layers that retain users within platform ecosystems.
Opt-out mechanisms based on text-and-data-mining exceptions remain fragmented, technically complex and difficult to enforce. In practice, refusing participation can reduce visibility even further.
Europe risks defending media freedom rhetorically while allowing the informational infrastructure supporting democratic debate to weaken economically.
The Forgotten Layer: News Agencies as Critical Infrastructure
One crucial element remains largely absent from public debate: the role of news agencies themselves.
News agencies are not ordinary B2B content suppliers. They constitute foundational verification infrastructure. They provide the first structured and verified data layers upon which newspapers, broadcasters, online media, institutions, governments and corporations build their own information environments.
During the European copyright reform process, agencies were formally recognised as beneficiaries of neighbouring rights. In practice, however, only a limited number of large agencies with direct consumer publishing operations captured meaningful revenues from these frameworks. Most agencies remained economically marginalised.
The European interpretation that only directly published consumer-facing content qualified for meaningful protection fundamentally underestimated the systemic role of agencies.
This misunderstanding has strategic consequences.
Europe already treats energy grids, telecommunications, cloud systems and cybersecurity capabilities as strategic infrastructure. Yet factual information systems, which underpin markets, institutions, compliance systems and democratic decision-making, are still largely treated as ordinary market products.
That distinction no longer reflects reality.
From Copyright Debate to Democratic Infrastructure
The discussion therefore needs to evolve beyond traditional copyright logic.
The core issue is not merely whether publishers receive compensation for snippets or summaries. The deeper issue concerns the resilience of democratic information infrastructures themselves.
At the World Press Freedom Day discussions organised by EANA in the European Parliament, this point became increasingly explicit. Journalism is not simply a creative sector competing for audience attention. Professional reporting, verification and agency infrastructure constitute part of the operational backbone of democratic societies.
EANA on World Press Freedom Day 2026: Safeguarding Press Freedom Means Protecting Democracies
Governments, corporations and institutions already rely on structured information feeds for compliance, geopolitical risk assessment, cybersecurity monitoring and operational decision-making. NIS2, cybersecurity governance frameworks and supply-chain risk obligations further increase this dependency on trusted signals.
Generative AI does not eliminate the need for verified journalism. It increases the strategic importance of verified information layers while simultaneously redirecting economic value toward synthetic interfaces.
The Australian Mirror
Australia offers one of the most important experiments in this debate.
Through the News Media Bargaining Code, Australia attempted to create mandatory negotiation frameworks between large technology platforms and publishers.
A recent Nieman Lab analysis explores both the successes and the limitations of the Australian model.
The proposed News Bargaining Incentive introduces a “pay or be taxed” logic under which platforms contributing insufficiently to journalism funding could face levies on local revenues.
Politically, this represents a significant shift. States increasingly recognise that purely market-driven mechanisms cannot sustain professional journalism ecosystems indefinitely.
Yet the Australian experience also exposes limitations. Larger publishers benefit disproportionately while smaller organisations and agencies remain comparatively weaker. More importantly, AI-driven answer layers risk remaining outside the scope of many existing frameworks.
Europe should study these lessons carefully before reproducing the same structural blind spots.
Toward an Opt-In Infrastructure Model
A sustainable future requires a transition toward explicit opt-in frameworks for professional journalistic content across all forms of AI usage.
This includes training, inference, AI summaries, answer generation, autonomous agents and synthetic news interfaces.
Machine-readable rights reservations, API-based licensing systems, watermarking standards and protocols such as MCP should form the technical basis for modern information governance.
The principle itself is straightforward.
If societies would never accept the unrestricted reproduction and monetisation of entire newspapers without permission, they should not accept functionally equivalent reuse inside AI-generated answer systems merely because the content has been fragmented and synthesised.
The issue is not technological progress. The issue is infrastructural sustainability.
Financing the Backbone
Any future European framework must also rethink how financial flows are distributed.
A purely market-driven allocation model rewards brand scale and negotiation leverage. It does not necessarily protect pluralism or infrastructural resilience.
A more coherent system would explicitly recognise the ecosystem role of news agencies as verification backbones while supporting publishers according to measurable public-interest criteria such as editorial capacity, local coverage and democratic relevance.
The logic is identical to other infrastructure sectors.
Europe already accepts that energy grids, telecommunications networks and cybersecurity capabilities require coordinated investment because markets alone do not guarantee resilience.
Professional information systems deserve the same strategic treatment.
Conclusion — Leaving the Information Wild West
The central challenge is now difficult to ignore.
News agencies are not historical relics or interchangeable content vendors. They are strategic infrastructure producing the verified facts, images and structured data required for democratic governance, institutional trust and informed decision-making.
Technology platforms themselves depend on this infrastructure. Their AI systems require verified information inputs to remain credible. Publishers depend on agencies to maintain cost-efficient international coverage and verification capacity. Governments depend on trusted information systems to resist disinformation and foreign influence operations.
Yet the current economic model increasingly weakens the very infrastructure upon which all these actors rely.
This is not a domain that can be left entirely to short-term advertising incentives, engagement metrics and fragmented market logic. Europe needs a coherent framework recognising our news agency infrastructure as a strategic asset in the AI era.
That framework must include explicit opt-in principles- machine-readable licensing systems with infrastructure-oriented funding mechanisms, AI transparency obligation, modern API-based governance and most of all the recognition of news agencies as critical informational infrastructure.
Without this shift, Europe risks allowing its democratic information backbone to erode while synthetic systems continue extracting value from increasingly underfunded professional journalism.
No ship navigates safely without reliable charts.
In democratic societies, those charts are drawn every day by journalists at news agencies.
