The European Commission and the Board of Digital Services Coordinators have published their second annual report on systemic risks under the Digital Services Act. Disinformation is not a marginal nuisance but a recurring systemic risk with real-world consequences, from electoral processes to public health. The DSA’s risk assessment framework, now in its second year, compiles insights from platforms’ risk assessments and civil society input, including the EFCSN’s.
The EFCSN contributed evidence to this year’s report, and our input is quoted in three of its chapters.
Some of the findings, in the report’s own words and ours:
Inauthentic content limits real people’s speech
The report quotes our submission directly: “By flooding communication spaces with falsehoods or inauthentic content, the freedom of expression of EU citizens is unfairly limited as their speech might be drowned out by inauthentic speech (e.g. from bots or sockpuppets).”
Disinformation is usually discussed as a content problem. It is also a freedom of expression problem for the citizens whose speech competes with inauthentic accounts. Additionally, by attacking users or public figures through false and misleading claims, Europeans may be discouraged to participate in public debates.
Virality pays, accuracy doesn’t
On civic discourse, the report cites our finding that “the monetization of polarizing and misleading content on social media is driven by engagement-based revenue models that reward virality rather than accuracy. These incentive structures favour emotionally charged and divisive narratives, which are systematically amplified through algorithmic recommendation systems.”
Research led by our member Science Feedback measured this difference. On YouTube, low-credibility accounts receive around 8 times more interactions than high-credibility accounts. On Facebook, around 7 times. The incentives also produce content at scale: our member Maldita.es identified 550 TikTok accounts that posted over 5,800 AI-generated videos of protests that never happened, made to drive engagement and qualify for TikTok’s monetisation programme.
Elections are targeted, and not only during campaigns
The report identifies foreign interference and information manipulation, including coordinated inauthentic behaviour on and off platform, as a systemic risk to civic discourse and electoral processes. Our members have documented how this works in practice. Funky Citizens found hundreds of bot-like accounts pushing identical messages during Romania’s 2025 presidential elections. Demagog traced a coordinated operation that targeted Moldovan diaspora voters through Facebook ads.
This work goes beyond debunking individual claims. European fact-checkers monitor the networks and incentive structures behind manipulation.
Health misinformation reaches patients
On public health, the report reflects our members’ evidence: “Health misinformation poses a threat not only to public health more broadly, and it carries steep economic costs.” The most recurring topics our members expose are cancer treatments, vaccines, their effectiveness, origin and contents and so-called wellness trends.
Generative AI is accelerating this. Our members have uncovered networks of AI-generated “doctors” giving medical advice and selling unproven supplements on TikTok, YouTube and Facebook. As seen in recent investigations, Science Feedback exposed AI avatars dispensing questionable advice on TikTok and impersonating real doctors on YouTube; the Pravda Association uncovered over 30 European profiles using fake experts to push unproven supplements; and Full Fact revealed a surge in Facebook accounts using synthetic doctors to market weight-loss patches.
The economic costs are measurable. Disinformation is not an abstract harm. It inflicts real, physical damage on people’s lives.
Scams are among the most prominent risks of the year
The report names financial scams as among the most prominent and recurrent systemic risks of the reporting period, spread through fake endorsements, misused official symbols and pressure tactics like fake countdown timers. Our members see the scale directly: Greece Fact Check alone reported over 4,000 unique scam links to Meta in nine months, most of them sponsored content.
Platforms are not just hosting this risk. They are earning from it.
What now?
For the second year running, the report does not name any mitigation practice as best or even good practice. It states that a measure being mentioned does not mean it works or is applied in practice. Everything platforms report remains self-reported.
At the same time, measures with evidence behind them are being reduced. Meta has severely slashed financial compensation for its third-party fact-checking partners while rolling back key support mechanisms. Google removed the fact-checking snippet from search results after it served fact-checks over 120 million times in six months in the EU alone.
The risks are documented in the regulator’s own report, partly with evidence from European fact-checkers. What we need now is more action from platforms, and enforcement if that action does not come.
