Mario Scharfbillig works at the Joint Research Centre (JRC) of the European Commission and co-authored the report Fractured Reality – how democracy can win the global struggle over the information space, published by the JRC. He notes: “My answers are based on the report and do not necessarily reflect the position or opinion of the European Commission. We provide independent, evidence-based knowledge and science to support EU policy making.”

---

In the report, you introduce the term ‘fantasy-industrial complex’. What do you mean by it, and what does it reveal about the way narratives are created and spread today?

Mario Scharfbillig: We use this term to describe a fundamental shift in how false and misleading information operates today, compared with ‘classic’ disinformation from the early 2000s and before. In the past, disinformation, for example about human-caused climate change, typically took the form of specific, identifiable falsehoods that were meant to be believed and that could be corrected once the facts became known and sufficiently widespread. People shared a common understanding that truth, at least in principle, was something that could be found.

Since then, societies have moved through a ‘shock and chaos’ phase marked by growing indifference to truth. This phase saw the rise of ‘alternative facts’, increasing bot activity, and FIMI (Foreign Information Manipulation and Interference), all of which have now blended into what we call the fantasy-industrial complex. We describe this as a self-organising, loosely coordinated ecosystem in which politicians, platform algorithms, corporate actors, various types of media (including influencers), and ordinary citizens all participate. Together, they blend kernels of truth with misrepresentation and outright falsehood to construct entire alternative versions of reality.

In this context, specific falsehoods aren't necessarily meant to be believed. Their function is often to distract, generate distrust, and normalise anti-democratic attitudes. As a result, common ground starts to look unreachable, and democratic debate increasingly appears challenging, if not futile. In short: it fractures realities.

---

How is the advent of Generative AI impacting an already challenging situation?

Mario Scharfbillig: We find that GenAI acts as an amplifier of pre-existing structural problems rather than an entirely new threat, though it does introduce some new dynamics. The most important are the ‘illusion of knowledge’, the integration of GenAI into social media platforms, and the scale and speed at which GenAI produces content.

The ‘illusion of knowledge’ relates to the fact that LLMs (large language models) rarely say ‘I don't know.’ They answer all questions confidently, creating a false sense that a question has been thoroughly addressed, even when it hasn't. We call this phenomenon ‘epistemia’, and it has the potential to foster false feelings of competence in users, something that research since then has increasingly confirmed. People appear more confident in their views after interacting with AI, even when the content they received was false or biased, which could further harden political debates.

Second, AI is now embedded directly in major platforms (e.g. Grok, Meta AI, Truth Search) and indirectly through the growing volume of AI-generated content, or ‘AI slop’, posted on them. This extends the attention economy's engagement-driven incentives to easily produced, personalised AI content, making the fracturing of realities described above even more pressing.

On top of that, the scale and speed of these tools enable large-scale, personalised, and persuasive content, including deepfakes, in almost any language and at low cost, within a system that does not prioritise accuracy. This is a recipe for further fracturing of common ground, though whether that will actually happen still needs further investigation.

---

An entire section of the report is devoted to strengthening Europe's digital sovereignty. What are its main recommendations, and why are they so important?

Mario Scharfbillig: We argue that even well-designed EU policies for countering disinformation and harmful business models may struggle to work in practice, because most of the platforms and infrastructure that make up Europeans' information space are owned outside the EU. We draw a distinction between ‘digital sovereignty’ (the legal right to set rules) and ‘digital autonomy’ (actual control over the technology itself), arguing that the EU has plenty of the former but maybe too little of the latter.

This dependency creates several risks: direct and indirect interference in democratic debate through platform algorithms’ ability to shift content priorities; the threat that platforms could go dark for EU users at a foreign owner's discretion; a highway of influence for foreign actors; and, through mass data collection, growing transparency of EU citizens to outside actors.

Our main proposals are intended to help the EU regain the initiative, some of which are already being acted upon:

1. Build European digital infrastructure through initiatives like EuroStack and DC-EDIC, and use the EU's annual public procurement to favour EU-owned systems.
2. Support decentralised social media built on open protocols – such as the FediVerse (e.g. Mastodon), ATmosphere (e.g. BlueSky), and the Eurosky project –  which can't be unilaterally shut down and which open the door to European ‘middleware’ innovation.
3. Develop European AI that respects EU values, with strong guardrails and societal-impact monitoring.

We also believe it would be worth investing in a ‘CERN for Data and Democracy’. This would give researchers faster access to platform data, allowing research and policy to keep pace with a fast-moving information environment. One such initiative by the Commission is currently open for application.

---

Quality information is a cornerstone of a healthy democracy, yet media consumption patterns are changing rapidly. What can the EU do in order to safeguard it?

Mario Scharfbillig: This is a tricky question, because the EU needs to respect citizens’ autonomy in what they want to consume and get information on, while at the same time ensuring they have meaningful access to high quality information. Research shows that consumption patterns are indeed shifting rapidly, although perhaps slower than for example in the US. Our report argues not to focus too much on the content level when it comes to ensuring a healthy information space, but rather on the structural incentives of the system.

Increasingly, information produced is consumed through the presentation of algorithms of large platforms, and people seem to feel comfortable with this passive mode of getting informed, a phenomenon that is sometimes called the ‘News finds me’. Thus, it is not enough to just look at what content is available, but also how it is presented to citizens. The challenge with this is that the underlying business model rewards engagement above all else, which structurally privileges negative, emotional and conflictual content, and has no regards for quality criteria such as whether facts are true.

Our recommendations therefore – apart from the sovereignty angle described above and issues of how to counter disinformation – focus especially on ways to either change the current business model or enable new business models to emerge, which would have better aligned incentives to promote quality content. This could also be done through taxing advertising or harmful content, broader interoperability requirements enabling competition in how content is presented, or new verification markets for content.

Related article