Does information sovereignty mean excluding foreign media?

No. Foreign reporting increases diversity, challenges domestic assumptions and supplies expertise a small market cannot reproduce alone. The problem is not foreign origin but dependency without visibility, alternatives or exit. A transparent foreign source can strengthen Swiss democratic knowledge, while a domestic conglomerate that removes independent editorial capacity can weaken it.

Why is news deprivation a sovereignty problem in Switzerland?

By 2025 roughly 46 percent of the Swiss population fell into the news-deprived category in the relevant research classification, about 25 percentage points more than in 2009. Direct democracy asks citizens to decide repeatedly on technically and constitutionally complex questions, so the epistemic burden is unusually high. A country can preserve press freedom and excellent journalism while losing informational reach—a bridge that exists but cannot be reached provides no mobility.

What should Switzerland do first?

Recognise the information environment as a national capability and measure the complete system: ownership, shared editorial production, discovery channels, advertising flows, critical infrastructure providers, news-deprived groups, recommender behaviour across Swiss languages and the visibility of Swiss sources in AI answers. An independent observatory with a measurement mandate—open methods, public data, no authority over lawful opinion—is the practical starting point, followed by ownership transparency, platform accountability, arms-length support for public-interest journalism and audited operational dependencies.

Who Owns Your National Conversation?

Q: What is information sovereignty?

It is the practical capacity of a democratic society to sustain independent knowledge production, understand and govern the systems that distribute information, preserve meaningful plurality, and retain the ability to intervene when concentration, opacity or dependency threatens collective self-government. The sovereign subject is the democratic system as a whole—citizens, journalists, researchers, institutions, courts and the legal order—not the government of the day.

Thirty years ago, the Swiss information landscape was easier to describe.

People read regional and national newspapers. They listened to radio. They watched Swiss television. Depending on the language region, they also consumed media from Germany, France or Italy.

The system was never neutral.

Editors made choices. Publishers had interests. Political leanings were visible. Public broadcasting reflected institutional assumptions. Some perspectives received more attention than others.

But the system was legible.

You could name the newspaper.

You could identify the editor.

You could understand who owned the printing press.

You could criticise the broadcaster, cancel your subscription, write a letter, change the channel or buy a competing paper.

Most importantly, large parts of the population still encountered a recognisable set of shared events.

People disagreed about what those events meant.

They did not necessarily inhabit entirely different realities.

That distinction matters.

Today, the Swiss national conversation is no longer formed primarily inside Switzerland.

It is produced through a layered system of domestic publishers, public broadcasters, foreign media, search engines, social networks, video platforms, messaging applications, advertising markets, cloud infrastructure and recommendation algorithms.

Increasingly, it is also mediated by artificial intelligence.

A citizen may learn about a federal vote through an SRF article surfaced by Google, a politician’s TikTok video, a German YouTube commentator, a screenshot in a WhatsApp group, an Instagram carousel, a podcast, a Telegram channel or an answer generated by ChatGPT.

All of these may concern the same political question.

They do not pass through the same standards.

They do not disclose the same incentives.

They do not offer the same possibilities for correction.

And they are not governed by the same country.

The channels changed.

The ownership structures changed.

The economics changed.

The incentives changed.

The systems deciding what becomes visible changed.

But the democratic process still assumes that citizens can form sufficiently informed judgements about a shared political reality.

That assumption deserves more scrutiny than it receives.

Because the central sovereignty question of the information age may no longer be:

Who owns the media?

It may be:

Who owns the conditions under which a country understands itself?

The infrastructure nobody calls infrastructure

Countries treat certain systems as critical because society cannot function without them.

Electricity.

Water.

Transport.

Telecommunications.

Financial infrastructure.

Healthcare.

Food supply.

Digital networks.

We maintain reserves. We regulate operators. We map dependencies. We test recovery plans. We worry about foreign control, single points of failure and the time required to restore essential services.

But democratic societies depend on another system that is rarely treated with the same seriousness.

Their ability to observe reality.

To investigate events.

To produce credible knowledge.

To distribute it.

To dispute it.

To correct it.

And to convert public understanding into legitimate collective decisions.

This is the cognitive infrastructure of a nation.

It is not one newspaper.

It is not the public broadcaster.

It is not social media.

It is the entire system through which events become knowledge, knowledge becomes public understanding, public understanding becomes political choice, and political choice becomes government action.

A simplified democratic feedback loop looks like this:

Events and material reality
        ↓
Observation, reporting and evidence
        ↓
Distribution and discovery
        ↓
Selection, ranking and recommendation
        ↓
Public attention and understanding
        ↓
Debate, participation and voting
        ↓
Government decisions
        ↓
Policy and implementation
        ↓
A changed material reality

This is a control system.

And like every control system, it depends on the quality of its sensors, the integrity of its signals, the design of its interfaces and the time available to detect and correct error.

A democracy does not require agreement at every stage.

It requires something more basic:

Enough independent observation to identify error.

Enough plurality to challenge interpretation.

Enough transparency to understand influence.

Enough shared factual ground for political disagreement to remain meaningful.

If the information entering the loop becomes systematically distorted, the democratic system does not immediately stop functioning.

Votes still take place.

Parliaments still meet.

Newspapers still publish.

Politicians still debate.

The forms remain.

But the feedback loop changes.

A society can continue making decisions while gradually losing the capacity to understand the conditions under which those decisions are being made.

That is how information sovereignty can disappear without a single dramatic moment of surrender.

What information sovereignty is not

Information sovereignty is an uncomfortable term because it can easily be misunderstood.

It does not mean that a country should control all information reaching its population.

It does not mean excluding foreign media.

It does not mean creating a ministry of truth.

It does not mean that governments should decide which political opinions are acceptable.

It does not mean replacing private platforms with a government-run national social network.

And it certainly does not mean that domestic ownership automatically produces trustworthy information.

A state that controls newspapers, broadcasters, advertising and distribution may possess enormous information power.

It does not possess democratic information sovereignty.

It possesses information control.

The difference is fundamental.

A useful definition would be:

Information sovereignty is the practical capacity of a democratic society to sustain independent knowledge production, understand and govern the systems that distribute information, preserve meaningful plurality, and retain the ability to intervene when concentration, opacity or dependency threatens collective self-government.

The sovereign subject is not the government alone.

It is the democratic system as a whole:

citizens,
journalists,
researchers,
public institutions,
civil society,
courts,
infrastructure operators,
educational institutions,
and the legal order connecting them.

This produces three distinct but connected forms of sovereignty.

Informational sovereignty

Who owns the channels, infrastructure, data and systems through which information moves?

Who can amplify, suppress, interrupt, observe or monetise those flows?

Editorial sovereignty

Can journalists and institutions decide what to investigate and publish without improper pressure from owners, advertisers, platforms or government?

Epistemic sovereignty

Can a society still establish sufficiently reliable shared facts to make collective decisions?

Epistemic sovereignty does not require political consensus.

Citizens can agree on the evidence and still disagree profoundly about what should be done.

But when they can no longer agree on what happened, which institutions produced the data, whether any source can be trusted or whether evidence itself is legitimate, democratic disagreement becomes something else.

It becomes competition between incompatible realities.

Switzerland’s information system did not simply become more diverse

There is a comforting story about the transformation of media.

Once, a small number of newspapers and broadcasters controlled information.

Then the internet arrived.

Gatekeepers disappeared.

Everyone gained a voice.

Information became democratised.

Parts of this story are true.

Citizens gained access to sources that would once have been unreachable. Politicians could communicate directly. Whistleblowers could publish evidence. Marginalised communities could organise. Independent creators could build audiences without owning printing presses or broadcast licences.

Foreign perspectives became easier to access.

Corrections could travel faster.

But the disappearance of old gatekeepers did not produce a world without gatekeepers.

It produced new ones.

Search engines decide what can be found.

Recommendation systems decide what appears.

Social networks decide what spreads.

App stores decide what can be installed.

Advertising systems decide what can be monetised.

Cloud and content-delivery providers decide what remains available.

Artificial-intelligence assistants increasingly decide which sources are retrieved, summarised, cited or ignored.

The gatekeeper did not disappear.

It became less visible.

A newspaper editor can be named.

A publisher’s ownership can be investigated.

A broadcaster can be challenged in public.

A recommendation system may change continuously, affect every user differently and remain inaccessible to citizens, researchers and national authorities.

The issue is not that algorithms are necessarily malicious.

The issue is that they exercise consequential power without equivalent observability.

A society can see the output.

It often cannot inspect the system that selected it.

That asymmetry is a sovereignty problem.

Many titles. Fewer voices.

Switzerland still has a substantial media landscape.

It performs well in international press-freedom comparisons. It retains strong journalistic institutions. Its major high-reach news organisations remain predominantly Swiss-controlled. SRG SSR still provides a common public-service layer across four language regions.

These are real strengths.

But visible diversity can conceal structural concentration.

In German-speaking Switzerland, the three largest high-reach media providers account for roughly three quarters of the market.

In French-speaking Switzerland, the share is closer to seven eighths.

Across Swiss news websites, the five largest groups account for around 73 percent of cumulative use.

Even these figures do not tell the full story.

Ownership concentration measures who owns the brands.

It does not measure how much independently produced journalism exists behind them.

That requires another metric: editorial concentration.

In the German-speaking press, the proportion of editorial articles appearing in at least two titles increased from around 10 percent in 2017 to 26.5 percent in 2024.

Within one major newspaper network, more than half of the analysed material was shared.

In national political reporting, shared content reached more than 40 percent.

Even opinion and commentary—content that should reveal distinct editorial identities—became significantly more centralised.

This does not mean shared newsrooms are automatically bad.

Consolidation can keep regional titles alive. It can fund specialists. It can prevent the total disappearance of reporting in markets unable to sustain several full editorial teams.

But it changes what a newspaper title represents.

A country may still have many mastheads while losing independent observation, local knowledge, editorial disagreement and competing investigative capacity.

Brand plurality is not necessarily source plurality.

And source plurality matters because independent systems do more than offer different opinions.

They detect each other’s failures.

Plurality is a resilience mechanism.

The business model moved outside journalism

The transformation is not only editorial.

It is economic.

For much of the twentieth century, newspapers controlled both journalism and the commercial infrastructure surrounding it.

They produced content, sold advertising, managed subscriptions and controlled physical distribution.

The internet separated these functions.

Publishers still carry the cost of reporting.

But global platforms increasingly control discovery, audience data and advertising markets.

In 2024, estimated Swiss revenues from online search and social-media advertising exceeded the measured advertising revenues of Swiss press, television, radio and online media combined.

That is a profound structural shift.

The companies producing local, cantonal, investigative and multilingual journalism no longer control much of the market through which attention is monetised.

The platforms do.

Publishers are therefore pushed toward paywalls, subscriptions, marketplaces, events, commercial services, automation and cost consolidation.

Some of these adaptations are necessary and innovative.

But they create new trade-offs.

Subscription models can support deeper reporting and stronger direct relationships with readers.

They can also place high-quality information behind socioeconomic barriers.

Advertising-funded journalism can remain broadly accessible.

It can also become dependent on traffic volume, emotionally charged content and platform distribution.

Foundation ownership can protect long-term public-interest journalism.

It can also create dependency on a small number of donors if governance is unclear.

Public funding can sustain multilingual, regional and civic reporting the market would not produce.

It can also become political leverage if allocation is not independent.

There is no perfect business model.

The sovereignty question is whether the system contains enough diversity that the failure or capture of one model does not determine the entire national conversation.

The foreign influence question is more complicated than it looks

Switzerland has always consumed foreign media.

German-speaking citizens watch German television and read German publications.

French-speaking citizens consume French media.

Italian-speaking citizens are exposed to Italy’s much larger media market.

This is not inherently a weakness.

Foreign reporting can increase diversity, challenge domestic assumptions and provide expertise that a small national market could never reproduce alone.

A sovereign information system does not exclude foreign voices.

It benefits from them.

The problem begins when exposure becomes dependency without visibility or alternatives.

A German newspaper has an identifiable editorial structure.

A French broadcaster can be analysed as an institution.

A social-media feed is different.

It may mix Swiss journalism, foreign reporting, political advertising, creator commentary, entertainment, institutional communication, anonymous accounts and synthetic content inside the same interface.

The user may not know why a particular item appears.

The person publishing it may not know how it will be distributed.

The Swiss state may not know how the ranking system behaves in German, French, Italian or Romansh.

The platform may not disclose enough data for independent researchers to find out.

Foreign information is not the core threat.

Unobservable distribution power is.

The relevant distinction is therefore not simply domestic versus foreign.

It is governable versus ungovernable dependency.

A foreign news source operating transparently may strengthen Swiss democratic knowledge.

A domestic media conglomerate that removes independent editorial capacity may weaken it.

A foreign cloud provider with tested portability and credible failover may be less dangerous than a nominally Swiss provider from which systems and data cannot be extracted.

Nationality matters.

But control, observability, plurality and exit capability matter more.

What the algorithms actually do

The public debate about recommendation systems is often disappointingly simple.

One side claims that algorithms manipulate elections and radicalise entire societies.

The other claims that they merely show users what they already want.

The evidence is more complicated.

Large experiments around Facebook and Instagram during the 2020 United States election found that switching users to chronological feeds substantially changed what they saw and how they engaged.

But the researchers did not detect corresponding short-term changes in political polarisation or political attitudes.

A later independent randomised experiment on X found that activating its algorithmic feed altered exposure in a different way.

It promoted more conservative material, reduced exposure to traditional news organisations and shifted some issue opinions in a conservative direction.

It did not significantly change party identification or affective polarisation.

Research on TikTok has identified partisan imbalances in political recommendations, but exposure asymmetry alone does not prove downstream changes in beliefs.

The responsible conclusion is not that recommendation systems have no political effect.

Nor is it that every feed directly determines how people vote.

The conclusion is more important:

Systems that select political exposure can alter what citizens encounter, yet their effects vary by platform, design, context, user and time.

That uncertainty strengthens the case for scrutiny.

It does not weaken it.

When an infrastructure layer is powerful and its effects are uncertain, the answer is not blind trust.

It is observability.

Independent audits.

Research access.

User choice.

Transparent political advertising.

Meaningful appeal mechanisms.

And the ability to compare algorithmic and non-personalised feeds.

A democracy does not need to dictate the correct ranking of information.

It needs the capacity to understand how ranking systems behave.

The quietest Swiss warning

The most serious threat to Swiss information sovereignty may not be foreign propaganda.

It may not be media ownership.

It may not even be recommendation algorithms.

It may be withdrawal.

The proportion of people classified as “news-deprived”—those with low news use across a narrow range of sources—has risen dramatically.

By 2025, this group represented roughly 46 percent of the Swiss population in the relevant research classification, around 25 percentage points more than in 2009.

The term should be used carefully.

It does not mean that almost half the population sees no information.

Many people encounter enormous quantities of content.

They may use social media daily.

They may watch videos, read posts, receive screenshots and discuss politics in messaging groups.

But exposure to content is not the same as sustained use of journalistic news.

Research associates news-deprived media repertoires with lower political knowledge, lower participation and weaker connection to democratic society.

Social media does not necessarily compensate for the absence of regular journalism.

This is especially significant in Switzerland.

Swiss citizens are not merely asked to select representatives every few years.

They vote repeatedly on technically, financially and constitutionally complex questions.

Direct democracy places an unusually high epistemic burden on the population.

It assumes that citizens can reach information, compare arguments, understand uncertainty and identify the interests behind political claims.

A country may therefore preserve press freedom while losing informational reach.

It may preserve excellent journalism while fewer citizens encounter it.

It may preserve formal democratic rights while the knowledge required to exercise them becomes unevenly distributed.

This is not primarily a censorship problem.

It is an infrastructure problem.

A bridge that exists but cannot be reached does not provide mobility.

Journalism that exists but no longer enters large parts of the public conversation cannot fully perform its democratic function.

Finland: resilience without isolation

No country controls its entire information environment.

Finland depends on global technology platforms.

Its commercial media market is concentrated.

Its public broadcaster faces political and financial pressure.

It is not an information-sovereign paradise.

But it demonstrates what resilience can look like.

Finland combines strong press freedom, high levels of trust in news, established public-service media, regional journalism and a relatively strong culture of paying for digital news.

Most importantly, it treats media literacy as a national and lifelong capability.

Its national policy does not reduce media education to teaching children how to identify fabricated stories.

It involves schools, libraries, cultural institutions, researchers, civil society, media organisations and public authorities.

The objective is broader:

To help citizens understand how media systems work, how claims are constructed, how sources can be assessed, how digital environments shape behaviour and how participation can remain informed.

This matters because information resilience cannot be installed during a crisis.

A government cannot suddenly create public trust during an election.

A school system cannot repair decades of weak source literacy when a disinformation campaign begins.

A public broadcaster cannot become legitimate overnight.

Independent local journalism cannot be reconstructed immediately after its economic foundations disappear.

Resilience is accumulated before it is needed.

Finland’s strength is not independence from foreign platforms.

It is the presence of counterweights:

strong institutions,
civic competence,
direct audience relationships,
public-service capacity,
independent journalism,
and a cultural expectation that information systems should remain accountable.

This is sovereignty as preparedness rather than isolation.

Hungary: domestic control without democratic sovereignty

Hungary demonstrates the opposite lesson.

Over the period beginning in 2010, the Hungarian information system became increasingly aligned with one political project.

Ownership transfers, regulatory changes, public-media control, state advertising and pressure on independent outlets reinforced one another.

Before the change of government in 2026, government-aligned actors were estimated to control more than 80 percent of the media environment.

Trust in news fell to exceptionally low levels.

Opposition politicians received minimal visibility in parts of prime-time public-service news.

Independent journalism did not disappear entirely.

But the system surrounding it became structurally hostile.

The Hungarian case matters because it destroys a simplistic assumption:

Domestic ownership does not guarantee information sovereignty.

A media system can become more nationally controlled and less sovereign at the same time.

If ownership, funding, regulation, public broadcasting and political power become aligned, the country may gain greater control over the information system while citizens lose independent access to knowledge.

The state becomes stronger.

The democratic system becomes weaker.

This is why information sovereignty must belong to the political community, not the government of the day.

Hungary also reveals another systems principle.

Capture is cumulative.

No single intervention needs to abolish press freedom.

A regulator becomes less independent.

State advertising becomes less neutral.

A broadcaster becomes more aligned.

An owner acquires more outlets.

Independent competitors lose revenue.

Local journalism weakens.

Access to public information becomes harder.

Each step may appear survivable.

Together, they form a self-reinforcing system.

And such systems are easier to capture than to rebuild.

A government can change after an election.

Trust, independent institutions, local reporting capacity and plural ownership structures take much longer to restore.

The Information Sovereignty Stack

Information sovereignty cannot be evaluated through one metric.

A country can rank highly on press freedom while becoming dangerously dependent on a few platforms.

It can have many media owners but one dominant distribution system.

It can have strong infrastructure but politically captured journalism.

It can produce excellent reporting that large parts of the population never see.

A more useful model is a stack.

1. Knowledge production

Does the country retain the capacity to observe and report its own affairs?

Are there journalists covering municipalities, cantons, courts, companies and national institutions?

How much reporting is original?

How much is syndicated?

How many independent editorial teams can investigate the same event?

2. Ownership and economics

Who can appoint leadership, withdraw capital or redirect editorial strategy?

How concentrated is audience power?

Which business models finance journalism?

How much revenue has moved from domestic media to global platforms?

Can new organisations enter the market?

3. Distribution and discovery

How do citizens encounter information?

Through direct visits?

Broadcasting?

Search?

Social media?

Messaging applications?

Video platforms?

Newsletters?

Artificial-intelligence assistants?

Can important public-interest information reach citizens without passing through one unavoidable intermediary?

4. Recommendation and moderation

Who determines what is amplified, suppressed or targeted?

Can users choose non-personalised feeds?

Can researchers inspect platform behaviour?

Are political advertisements archived?

Do moderation systems work equally across national languages?

5. Infrastructure and continuity

Where are news systems hosted?

Which DNS, cloud, CDN, analytics, identity and payment providers do they depend on?

Can media organisations migrate?

Can they continue publishing during a major outage, cyberattack, commercial dispute or platform suspension?

Are archives exportable and independently preserved?

6. Law and institutional governance

Are regulators independent?

Is ownership transparent?

Can journalists protect sources?

Is public-service media insulated from political capture?

Are government subsidies and state advertising allocated through transparent, arms-length processes?

Do courts provide meaningful redress?

7. Epistemic capacity

Can citizens identify sources?

Can they distinguish reporting from commentary, advertising and political persuasion?

Do they understand personalised feeds?

Can they evaluate evidence and uncertainty?

Do they retain enough trust to use credible institutions without becoming blindly deferential?

The stack matters because weakness compounds.

High-quality journalism cannot serve democracy if it is economically unsustainable, algorithmically invisible, technically unavailable or consumed by too few citizens.

Resilient infrastructure cannot create sovereignty if the content layer is captured.

Media literacy cannot compensate for the disappearance of independent reporting.

Ownership diversity cannot compensate for dependency on one platform.

A sovereign information environment requires plurality across the stack.

Artificial intelligence changes the question again

The current debate still focuses heavily on social media.

But the next transformation has already begun.

People are moving from searching to asking.

A search engine presents links.

An AI assistant presents an answer.

That difference is not cosmetic.

Search allows the user to see competing sources, even when rankings influence which ones appear first.

An assistant synthesises those sources into one response.

It decides what to retrieve.

What to exclude.

What to paraphrase.

What to cite.

What to describe as uncertain.

And what to present as settled.

A Swiss study using more than one thousand news-related prompts and thousands of cited sources found that AI-generated answers still contained partly or wholly false information in a meaningful minority of cases.

It also found significant differences in which sources became visible depending on the system, language and topic.

For Switzerland-related questions, Swiss media represented a substantial but far from complete share of sources.

For international topics, they nearly disappeared.

This creates a new distribution market.

A Swiss publisher may produce accurate, original reporting and still become invisible if AI systems do not index, retrieve, licence or cite it.

A language region may receive different answers because the available source ecosystem differs.

A correction may be published but fail to propagate into models or retrieval systems.

A source may be summarised without the context that made its reporting meaningful.

The future sovereignty question is therefore not only:

Who trains the models?

It is also:

Who decides which national knowledge enters the answer?

Who can inspect the retrieval process?

Who can correct the system?

Which sources remain economically viable when users no longer visit them?

And what happens when a foreign AI assistant becomes the primary interface through which citizens understand domestic politics?

The next generation may not read a newspaper.

It may not use Google.

It may simply ask.

Whoever answers becomes part of the cognitive infrastructure of the country.

What Switzerland should not do

The risks are real.

That does not make every proposed solution wise.

Switzerland should not build a government-controlled national social network and assume citizens will use it.

It should not exclude foreign media.

It should not impose blanket domestic-hosting rules that create a fragile Swiss monoculture.

It should not force publishers into one funding model.

It should not give authorities broad powers to determine political truth.

It should not require “algorithmic neutrality”, because no neutral ranking of millions of items exists.

It should not attach subsidies to favourable reporting or vague administrative judgements of quality.

And it should not confuse public scepticism with democratic health.

A population that trusts nothing is not sovereign.

It is paralysed.

The objective is not unlimited control.

It is plural, observable and recoverable interdependence.

That is the same logic that applies to cloud infrastructure, energy systems and supply chains.

Dependency is not failure.

Dependency without visibility, alternatives or exit options is.

What Switzerland can do

The first step is to recognise the information environment as a national capability.

Not a propaganda domain.

Not merely a media market.

Not merely a cultural policy issue.

A democratic capability.

Measure the complete system

Switzerland already measures aspects of media reach, ownership and journalistic quality.

But the layers remain fragmented.

The country should be able to answer:

Who owns the major media brands?
Which titles share editorial production?
Which platforms determine most news discovery?
Which advertising systems extract the revenue?
Which cloud, CDN, DNS and identity providers are operationally critical?
Which groups of citizens no longer encounter regular journalism?
How do recommendation systems behave across Swiss languages?
How visible are Swiss sources in AI-generated answers?

An independent Information Sovereignty Observatory could integrate these questions without becoming a regulator of opinion.

Its mandate should be measurement.

Open methods.

Public data.

Trend detection.

No authority to declare lawful political views true or false.

Make control visible

Citizens should be able to see the beneficial ownership, voting rights and editorial-control structures behind high-reach media.

Shared newsrooms and content networks should be transparent.

Political and issue advertising should be searchable at the advertisement level, including sponsor, expenditure, targeting criteria, dates and reach.

The objective is not to prohibit influence.

It is to make influence observable.

Demand platform accountability without prescribing speech

Large platforms and search engines should provide meaningful information about their recommendation systems, moderation processes and political-advertising activity.

Accredited researchers should receive privacy-protective access to data needed to audit systemic effects.

Users should have accessible alternatives to personalised ranking.

Appeal mechanisms should work in Swiss languages.

Material changes to recommendation systems should not remain entirely invisible to the societies affected by them.

Protect economic plurality

Public-interest journalism will not survive on moral praise.

Local, cantonal, investigative and multilingual reporting creates democratic value that advertising markets increasingly fail to finance.

Support can be justified.

But it must remain arms-length, formula-based, transparent and open to new entrants.

Government should fund capabilities, not political alignment.

Original reporting.

Local presence.

Accessibility.

Multilingual provision.

Investigative capacity.

Public archives.

No content approval.

No loyalty test.

Build direct channels

Media organisations should reduce their dependence on personalised platforms by strengthening websites, newsletters, podcasts, RSS, events, libraries and other direct relationships.

Public institutions should publish primary evidence in durable, accessible formats.

Citizens should be able to reach important information through several independent channels.

No country should discover during a crisis that its public communication depends on one foreign platform account.

Audit operational dependencies

High-reach media and public institutions should understand which cloud, DNS, CDN, app-store, analytics and identity systems they rely on.

They should test migration.

Backup publication.

Archive restoration.

Emergency communications.

And provider failure.

Sovereignty is not where the server stands.

It is whether the organisation can still act when the provider, contract or political environment changes.

Treat media and AI literacy as lifelong infrastructure

Switzerland does not need one centralised national curriculum.

Its federal system is capable of more nuanced implementation.

But the capability itself should be national.

Children need source recognition, evidence evaluation and recommender awareness.

Adults need practical skills for evaluating referendum claims, political advertising and synthetic media.

Older citizens need support against impersonation, fraud and manipulated content.

Journalists and public officials need standards for transparent AI use, provenance, corrections and human accountability.

The goal is not to teach citizens what to think.

It is to preserve their capacity to determine why they believe something.