Customize your cookie preferences

We respect your right to privacy. You can choose not to allow some types of cookies. Your cookie preferences will apply across our website.

We use cookies on our site to enhance your user experience, provide personalized content, and analyze our traffic. Cookie Policy.

Back to HubBack to Compass
Trusted AIAI GovernanceSovereign AIEuropePublic Infrastructure

AI Has a Trust Problem – Europe's Public Broadcasting Model May Hold a Clue

11 min read

A thought that stayed with me after TEDxBerlin.

I recently had one of those moments where an idea from a completely different field suddenly clicked into place.

At TEDxBerlin, Aljoscha Burchardt mentioned that the German public broadcasting model could act as a guiding model for building an AI ecosystem that is not built only around commercial interests, but also around serving the public good.

That stayed with me. It has kept my head running for over a week now.

Not because public broadcasting and AI are the same thing. They are not.

But because the comparison raises a structural question that feels highly relevant to AI right now:

How do you build powerful technology infrastructure in a way that does not only optimize for market dominance, but also for trust, representation, accountability, and societal benefit?

When people talk about Europe's AI future, the question is often framed like this:

Where is Europe's OpenAI?

It is an understandable question. Europe needs competitive companies. It needs serious technical capability. It needs compute, chips, cloud infrastructure, models, data, tooling, talent, and capital.

But I keep wondering whether the framing is still too narrow.

Maybe the question is not only whether Europe can produce one dominant AI platform.

Maybe it is also whether Europe can build an AI ecosystem that people, companies, institutions, and governments can actually trust.

Public Broadcasting as a Trust Architecture

Across Europe, public broadcasters were not created simply to compete in the media market.

They were created with a broader mandate: universal access, independent information, cultural representation, linguistic diversity, education, democratic debate, and public trust.

Each country has its own institutions, legal context, funding model, political tensions, and cultural identity. ARD and ZDF are not the BBC. France Télévisions is not Rai – Radiotelevisione Italiana. Yle is not Radiotelevisión Española.

And yet, these organisations are part of a broader European public media tradition.

The important point is not that public broadcasting is perfect. It is not.

Public broadcasters face political pressure, funding debates, accusations of bias, digital disruption, and real questions about relevance in a platform-driven media environment.

But the model still matters because it shows something important:

Trust does not appear by accident.

It is built into institutions, mandates, standards, safeguards, and accountability structures.

Public broadcasting is trusted when it works because it is built around specific trust mechanisms:

  • Editorial independence.
  • Professional journalistic standards.
  • Representation of different regions, languages, and communities.
  • Accountability to the public.
  • A mandate beyond commercial attention.
  • Broad access, not only profitable audiences.

That is the part I find interesting for AI.

Because AI trust will not come from capability alone.

It will come from the systems around capability.

Where the EBU Fits In

This is also where the European Broadcasting Union (EBU) becomes useful as an analogy.

Many people may not know the EBU by name. If they know it at all, they may know it as the organisation behind the Eurovision Song Contest.

But the EBU is much more than Eurovision.

It is an alliance of public service media organisations. According to the EBU, it brings together 113 member organisations in 56 countries. Its members are broadcasters and media organisations with public service mandates.

Its role is not to replace national broadcasters.

It does not turn ARD, BBC, France Télévisions, RAI, Yle, RTVE and others into one single European broadcaster.

Instead, it creates a coordination layer around them.

That coordination includes areas such as news exchange, content collaboration, sports rights, technical standards, innovation, training, legal and policy work, and public media strategy.

In simple terms: the EBU helps public media organisations cooperate across borders while still keeping their national mandates, languages, audiences, and institutional identities.

That distinction matters.

Because it is very different from saying "Europe needs one central institution to do everything."

The more interesting idea is this:

Europe sometimes builds strength by coordinating many public-interest actors around shared standards, infrastructure, and trust mechanisms.

The AI Question Is Bigger Than Models

This is where the analogy becomes relevant to AI.

The European AI challenge is not only a model challenge.

It is a broader provider and infrastructure challenge.

Models matter, but they are only one layer. Europe also needs capacity across chips, compute, cloud, data infrastructure, deployment environments, cybersecurity, evaluation, tooling, integration, sector-specific applications, governance, and procurement.

If we reduce the question to "Where is Europe's OpenAI?", we risk simplifying the problem too much.

The real question is broader:

What kind of European AI ecosystem would allow companies, public institutions, and citizens to use AI with confidence?

Today, many organisations want to use AI but are unsure how to do so responsibly.

This is especially true for public institutions, regulated companies, cities, hospitals, schools, broadcasters, and infrastructure providers.

They are not only asking:

Which model performs best?

They are asking:

  • Who controls the capability?
  • Where does the data go?
  • Can we explain the system?
  • Can we audit it?
  • Can we switch providers?
  • What happens if access, pricing, regulation, or permitted use changes?
  • Can we use this inside critical workflows without creating dependencies we do not understand?

These are not abstract policy questions.

They are implementation questions.

And implementation is where trust either becomes real or collapses.

What Would Trust Mechanisms for AI Look Like?

This is where the public broadcasting analogy becomes productive.

If independent journalism is one trust mechanism for broadcasting, independent model evaluation may become one trust mechanism for AI.

If editorial standards help define responsible broadcasting, model governance standards will help define responsible AI behaviour.

If representation matters in public media, representative and multilingual data matters in AI.

If public broadcasters are accountable to society, AI systems used in public or critical contexts need accountability to users, citizens, regulators, and affected groups.

If public broadcasters have a mandate beyond maximizing attention, AI systems used in public-interest contexts need objectives beyond engagement, automation, or lowest-cost deployment.

And if the EBU helps coordinate public media without replacing national broadcasters, Europe may need coordination layers that help organisations adopt AI without each building the entire trust stack alone.

That trust layer could include things like independent evaluation, multilingual benchmarks, procurement frameworks, audit practices, governance templates, and fallback options.

This does not mean AI should be built like public broadcasting.

But it does suggest that Europe's AI ecosystem needs more than isolated products and fragmented pilots.

It needs shared trust infrastructure.

The Funding Question Cannot Be Avoided

There is also an uncomfortable but important point here.

If Europe wants true European capability, it will have to build real European infrastructure.

That means not only frameworks, principles, and policy papers.

It means actual capacity: compute, cloud, chips, data infrastructure, foundation models, sector-specific tools, deployment environments, security layers, and implementation partners.

And that requires money.

A lot of it.

Public broadcasting offers a useful analogy for public mandate and coordination, but AI infrastructure is capital-intensive in a different way. It is unlikely that public institutions or taxpayer money alone can fund everything Europe needs.

At the same time, leaving the ecosystem entirely to commercial incentives creates its own risks.

The challenge is not public versus private.

It is how to balance commercial investment with public benefit.

This is also where the EBU becomes the more useful analogy than one single national broadcasting system.

A national public broadcaster can show how trust, public mandate, and accountability can be institutionalised locally. But AI infrastructure is too capital-intensive, too technical, and too globally competitive to be solved country by country alone.

If Europe wants to be truly competitive, it will need ways to pool effort without erasing national needs.

That means shared infrastructure, shared standards, shared evaluation methods, shared procurement pathways, and possibly shared financing vehicles – while still allowing different countries, sectors, and institutions to serve their own contexts.

In that sense, the lesson is not that Europe needs one central AI institution.

It may be that Europe needs stronger coordination layers that allow national, public, private, and institutional actors to build together at a scale none of them could reach alone.

Europe will likely need hybrid models: public funding, private capital, institutional procurement, shared infrastructure, anchor customers, strategic industrial policy, and commercial providers that can still operate within trusted frameworks.

This is a difficult balance.

But it may be the balance that matters most.

Because if Europe wants an AI ecosystem that serves society, it cannot only ask who can build the most powerful system.

It also has to ask who funds the infrastructure, who governs it, who benefits from it, and who remains accountable when it becomes part of critical workflows.

From Public Media to Public-Interest AI

The public broadcasting analogy is not a blueprint.

It is a lens.

It helps us see that trust is not just a feeling. It is a design problem.

I do not think the public broadcasting analogy gives us a ready-made blueprint. But it may give us a useful starting point for asking better questions about trust, coordination, and public-interest infrastructure in AI.

In broadcasting, trust is supported through independence, standards, representation, accountability, and public mandate.

In AI, the equivalent trust mechanisms will look different.

They may involve monitored training processes, independent evaluations, transparent documentation, representative data, public-interest benchmarks, clear liability structures, robust procurement standards, and credible alternatives to foreign dependencies.

They may also require coordination across countries, sectors, companies, research institutions, public agencies, and infrastructure providers.

No single actor can do all of this alone.

And that is exactly why the analogy matters.

The European public broadcasting model shows that it is possible to build national institutions with distinct mandates while also coordinating around shared infrastructure and public-interest principles.

The AI ecosystem may need something similar, but adapted to a very different technological and economic reality.

The Questions We ALL Should Be Asking

So yes, Europe should ask whether it has enough AI companies, model providers, compute capacity, cloud infrastructure, chips, talent, and capital.

Those questions matter.

But they are not enough.

Because the deeper question is not only whether AI can be built in Europe.

It is whether AI can be trusted in Europe.

  • Trusted by companies.
  • Trusted by public institutions.
  • Trusted by citizens.
  • Trusted inside critical workflows.
  • Trusted across languages, sectors, and national contexts.

That trust will not come from one model launch.

It will come from the ecosystem around the technology.

Maybe the question is not only:

Where is Europe's OpenAI?

Maybe it is also:

  • What would Europe's trusted AI adoption infrastructure look like?
  • Who is building the coordination layer that makes it possible?
  • And how do we fund it in a way that balances commercial ambition with public benefit?

And let's be clear on one thing – these questions are not unique to Europe.

Any nation, region, or economic bloc that wants to adopt AI at scale will face a similar challenge.

Europe, India, Japan, Brazil, African regional blocs, Southeast Asia, and others – hopefully even the US – will all approach the question differently. Their institutions, markets, languages, legal systems, and political priorities will not be the same.

The question for us all is hopefully this one:

Which nation, region, or economic bloc will build the most trusted AI ecosystem first?

Because the next phase of AI competition may not be won only by whoever builds the most powerful model.

It may be won by whoever creates the conditions for AI to be adopted safely, confidently, and at scale.

That means capability, yes.

  • But also trust.
  • Governance.
  • Infrastructure.
  • Funding.
  • Public legitimacy.
  • And practical adoption pathways for real organisations.

Europe has a real opportunity here.

Not because it can simply copy the platform race.

But because it can help define a different one:

the race to build AI systems that societies can actually trust and use.

Sources and Further Reading