Identifying Disinformation: Conference Recap

On 24 April 2026, we organised a conference in collaboration with the European Parliament Liaison Office in Malta, the University of Malta (UM), and the Malta College for Arts, Science and Technology (MCAST).

Discussions on disinformation often focus narrowly on false content online. Yet the conference repeatedly returned to a broader point: disinformation is not simply a problem of misleading posts or malicious actors, but of economic incentives, weakened institutions, fragmented publics, and rapidly changing technological systems.

The five takeaways below reflect that wider perspective.

Take Away 1: The Attention Economy Rewards Outrage over Accuracy, and Spectacle over Expertise

For many of the speakers present, once optimistic about the internet in its early days, the current moment represents a painful reversal.  A decade ago, the working assumption was that social media would democratise access to information and hold power to account. The Arab Spring seemed to confirm this. The pendulum has since swung hard in the other direction, and understanding why matters.

Digital platforms are advertising businesses, and the commodity they sell is attention. The content that attracts the most attention is not the most accurate; it is the most emotionally charged. Outrage, fear, and tribal solidarity drive engagement, and engagement drives revenue. Platforms are structurally incentivised to amplify inflammatory content and structurally indifferent to whether it is true. Disinformation is, before anything else, a phenomenon of the business models we use to provide critical infrastructure.

Engagement not only amplifies inflammatory content; it also reshapes what counts as authority. The issue was framed through Max Weber’s three bases of authority: traditional authority, rooted in customs, habits, and social structures; legal-rational authority, grounded in credentials and law; and charismatic authority, based on personality and spectacle. We are witnessing the simultaneous collapse of the first two. The algorithm does not care about qualifications; it measures engagement. The influencer with two million followers and no relevant knowledge is, by the metrics that now govern public attention, more visible than the specialist with decades of research behind them. Authority has become an iterative game, no longer residing in the speaker but conferred by the audience moment by moment. Expertise has become difficult to distinguish from performed expertise.

Take Away 2: The Fourth Estate is in Deep Trouble

The popular narrative is that journalism is the institution best placed to serve as the fourth estate, the press and news media understood as an unofficial democratic power that holds authority to account, informs citizens, and sustains public deliberation. That institution is now under pressure on multiple fronts.

In Malta, the pressure is especially clear. 85% of Maltese respondents now use social media or online platforms to follow current affairs, the highest figure in the European Union and eleven points above the bloc average.

The same European Parliament survey finds that 45% of Maltese respondents encounter disinformation often or very often, yet only 14% report any lack of confidence in their ability to recognise it. Malta, therefore, combines the most platform-dependent news environment in the Union with one of the most self-assured populations when it comes to telling reliable information from unreliable.

These patterns sit within a broader restructuring of the news economy. Approximately 65% of digital advertising now goes to Facebook and Google, leaving traditional outlets competing for a shrinking share alongside influencers and other forms of infotainment. At the same time, journalists are increasingly forced to play the algorithm’s game, which means constantly minding the threat of algorithmic suppression. European news outlets have seen their visibility dramatically reduced for weeks or months after publishing content that automated content classifiers flagged as biased, effectively removing their reach without sanction or right of appeal. The cumulative effect is that running a newsroom is becoming commercially unsustainable just as the demands placed on the function it performs are growing.

The consequences for a small media market like Malta’s are stark. Maltese-language journalism has largely collapsed as a commercial proposition, and there is no Maltese journalist currently based in Brussels covering the institution that produces legislation Malta is obliged to transpose into national law.

The function being eroded is a specific one. Trained journalists bring verification practices, editorial oversight, legal accountability, and the institutional capacity to investigate wrongdoing and sustain a shared factual record. Podcasters and influencers contribute differently, operating outside institutional constraints, broadening participation, and reaching audiences that legacy media often miss. Both have a place in the contemporary public sphere, but the roles they perform are not interchangeable, and a healthy information environment depends on preserving the conditions under which both can do their distinctive work.

In a hyperlocal media market like Malta’s, the interdependence is particularly visible. Stories researched by legacy outlets stimulate discussion that is then picked up and amplified by commenters and influencers as it moves through the ecosystem into more niche platforms. New details surface in that secondary circulation, which in turn feeds follow-up reporting by the original outlet. The relationship is symbiotic, yet it was hard to shrug off the sense at the conference that journalists felt influencers and big tech were eating their lunch. The symbiosis only works while the upstream function, the original reporting, remains commercially and institutionally viable.

Take Away 3: Disinformation Erodes Shared Truth and Spreads Through Real Bonds

One of the most important things to grasp is that disinformation need not involve outright fabrication. The BBC’s Panorama documentary on Donald Trump, broadcast in October 2024, is a recent illustration. The programme spliced together two sections of Trump’s speech of 6 January 2021, delivered nearly an hour apart, in a way that gave the impression he had directly called on his supporters to march on the Capitol and “fight like hell”. The BBC chair later conceded this was an “error of judgment”, and the director-general and head of news both resigned. The footage was real and unaltered; the editing did the work. Private information released at a strategically damaging moment, or accurate content stripped of its context, can be just as effective as an invented claim. The truth itself can be weaponised.

One of the most effective disinformation strategies discussed was “firehosing of falsehood”: deliberately flooding the information space with a high volume of contradictory claims at speed. The goal is not to persuade but to exhaust, and the broader effect is to destroy the common ground on which democracy depends. When people are overwhelmed with contradictory information, they stop trying to form accurate views and retreat into whatever confirms what they already believe. The endpoint of this strategy is not disagreement about values but about facts: when two halves of the same public cannot agree on what is real, the common ground democracy requires has already been eroded. 

Disinformation does not operate in a social vacuum. Narratives are tailored to find emotional hooks in specific cultural, economic, and social conditions, exploiting existing political loyalties, religious identities, and local grievances. Echo chambers and epistemic bubbles exist in political party clubs and family group chats as much as online. The disinformation actor does not need to implant a foreign belief; they need only to inflame a pre-existing sentiment, using the channels of trust that real social bonds provide.

Take Away 4: The Target Is No Longer Just Human Audiences, but AI Systems Themselves

Generative AI has shifted the terrain in two distinct ways. The first concerns how AI-generated content reaches human readers. A key concern discussed was the illusion of knowledge created by AI-generated text, in which outputs mimic the tone and format of authoritative sources convincingly enough that their professional appearance bypasses critical scrutiny. Users receive a false sense of having understood something they have not actually interrogated.

The second is more structural. A technical presentation made clear that the audience for some disinformation is no longer human at all. The Russian “Pravda Network“, an influence operation spanning fifteen European hubs, including Malta, created websites that attracted almost no human readers. Their target was the algorithm. When AI systems scraped those sites as training data, the bias was embedded in future model outputs at scale. Pseudo-expert content produced for engagement, scraped indiscriminately, becomes part of the substrate on which future AI responses are generated.

The implication is significant. The target of contemporary disinformation is no longer just public opinion but the epistemic infrastructure on which public opinion is increasingly formed. As more people consult AI systems as a first port of call for information, contaminating those systems upstream becomes a more efficient strategy than persuading individuals downstream.

Take Away 5: The Response Must Match the Scale of the Problem

No single intervention is sufficient. The most immediate priority is enforcing the Digital Services Act, and within it, a greater focus needs to go towards applying the risk assessments that the legislation already contains. It is increasingly clear that the algorithms platforms deploy create perverse incentives and tangible harms, and the risk assessment regime is the mechanism through which that evidence base is built.

Enforcement, however, cannot work if Europe lacks the digital sovereignty to back it. Between 80 and 90 percent of the EU’s post-COVID digital investment went to non-European hyperscalers, and digital resilience now needs to be treated as a strategic priority equivalent to food security and defence. The EU cannot regulate Big Tech into compliance while remaining entirely dependent on it.

These structural conditions create the space for the reconstructive work. Journalism needs new business models, and Europe should be actively investing in them. The Digital Fairness Act, currently being negotiated, faces a structural trap: heavy compliance burdens entrench the very platforms they are meant to discipline, while softer rules leave their incentives intact. The Act needs to be ambitious enough to bite and pragmatic enough to be applied. Media literacy belongs across entire educational curricula rather than being confined to optional modules, and beyond formal education, it needs to reach older adults, who are among the most exposed and least served demographics, through community infrastructure.

We have some questions

A recurring theme throughout the conference was that disinformation is not an isolated pathology of the internet age but a structural feature of contemporary information systems. It thrives where economic incentives reward outrage, where institutions capable of verification are weakened, and where fragmented publics no longer share common epistemic ground.

The challenge is therefore larger than fact-checking individual claims. It concerns the conditions under which societies are able to produce trusted knowledge at all. As AI systems become increasingly central to how information is accessed and interpreted, the stakes become even higher. The question is no longer simply whether citizens can distinguish truth from falsehood, but whether democratic societies can preserve the institutions, infrastructures, and cultures necessary for shared reality itself.

The value of a conference is not only in learning new information but also in thinking of new questions that need answering. These are the questions that remained with us most following the conference, which we will focus on answering in the coming weeks!

• Did we ever really share a common set of facts, or are we just romanticising the days when three TV channels told everyone the same story?
• Can we curb the worst of social media without handing power back to the old editors and broadcasters who used to decide what everyone saw?
• Do EU rules aimed at American tech giants actually help European rivals emerge, or do they slow those rivals down too, since the same compliance costs land on the smaller European challengers?
• Can journalism be paid for at the scale a democracy needs without becoming beholden to the governments or companies it is meant to hold to account?
• The same friendships and group chats that pass on reliable information are also how lies spread fastest. How do we strengthen the trust without strengthening the lies that travel alongside it?