Dear Disinfo Update readers,

As you read this newsletter, we’re working at full speed to finalise #Disinfo2026, which will take place next week in Vilnius, Lithuania. We’re expecting a full house, with more than 650 members of the counter-disinformation community, and highly relevant debates on everything from the latest developments in AI to new solutions and enforcement strategies. This year, we’ve also received a record number of registrations from national regulators, who are eager to learn more about what they can do and what they can learn from one another.

This comes at a key moment. Every week, we see new court cases and complaints filed at national level, showing how platforms and perpetrators can be held accountable for enabling and designing information operations. Together, these actions are building a legally robust body of evidence that should be the basis for quick and strong enforcement at EU level, finally delivering on the promises of the Digital Services Act.

If you can’t be with us next week, make sure to join our newly launched DisinfoHub, where additional resources, from conference materials to webinars, will be made available in the coming weeks.

We look forward to seeing you very soon!

NATIONAL RESPONSES 

Member state courts and regulators are pursuing accountability on several fronts, from platform liability to foreign-interference networks.

  • German court rules Meta liable for fake ads on Instagram andFacebook, rejecting Meta’s claim it acts as a passive host. (Reuters)
  • In the Netherlands, WHAT TO FIX files a DSA complaint over Meta’s failure to act on monetised pages linked to EU-sanctioned entities. (WHAT TO FIX)
  • French courts investigate a British consultancy executive over covert lobbying for a former Kazakh PM and Qatar-linked networks. (Le Monde)

Governments are strengthening their own defences, from election security and counter- disinformation units to media funding and expanded intelligence powers.

  • Sweden protects its elections through paper ballots with manual counts and a coordinated police-intelligence-electoral network. (Demócrata)
  • UK launches National Centre for Information Defence to counter Russian disinformation and deepfakes. (The Guardian)
  • France drafts bill requiring big advertisers spending over €1m a year to direct up to 10% of their digital ad budgets to recognised news outlets, or face a €75,000 fine. (Contexte)
  • Sweden, Germany, the Netherlands, and Austria expand spy agency powers amid Russian and Chinese threats. (Politico)

 

REGULATORY ENFORCEMENT AND PLATFORM GOVERNANCE

EU policy debates are intensifying around surveillance and AI accountability gaps.

  • The Digital Omnibus’s plan to move device-tracking rules from ePrivacy into the GDPR risks widening the exceptions that feed commercial and state surveillance, an op-ed argues. (Tech Policy Press)
  • AI Forensics finds a DSA/AI Act coordination gap leaves ChatGPT’s search function without proper risk oversight. (DSA Observatory)

Meta’s shift from professional fact-checking to Community Notes keeps drawing scrutiny.

  • Meta has begun testing Community Notes across 16 Latin American countries, its clearest step yet toward phasing out professional fact-checking in the region. (Reuters Institute)
  • Community Notes reaches just 3% of flagged Instagram posts, versus 20.5% on X, undercutting Meta’s claim the tool has “doubled” fact-checking coverage. (Indicator)
  • Meta’s Oversight Board warns that Community Notes cannot replace professional fact-checking, citing “real-world harms” under repressive conditions. (Oversight Board)
  • Study finds Community Notes curbs one-off violators, but repeat offenders post more, not less, after being corrected. (arXiv) 

Media and platform regulation is being contested from both sides, with platforms fighting rules and critics demanding tougher enforcement.

  • Meta launches legal challenge against UK media regulator over Online Safety Act, in a move critics say aims to delay the law’s implementation. (The Guardian)
  • An op-ed calls on French regulator Arcom to strip broadcast licences from CNews and Europe 1 over repeated pluralism breaches. (Le Monde)
  • An analysis challenges Australia’s “algorithm opt-out” proposal, arguing reverse-chronological feeds simply replace one algorithm with another. (Capital Brief)

 

CIVIL SOCIETY AND DEMOCRATIC RESILIENCE

New initiatives are emerging, but funding for disinformation research is running dry.

  • Crédit Mutuel’s regional press group EBRA is launching a think tank on “information quality and democracy,” co-chaired by Sophie Gourmelen and former CFDT chief Laurent Berger. (L’Informé)
  • A structural EU funding gap threatens civil society’s disinformation-research capacity, which has lost more than half its funding as US support collapses. (DFRLab)

Press freedom and independent verification are under growing pressure.

  • Major TV networks refuse to replace CNN in White House pool coverage following a joint First Amendment lawsuit against access bans. (CNN)
  • Bellingcat, the NYT, and other OSINT practitioners discuss why verifying war-crime evidence is getting harder, with AI fakes, shrinking satellite access, and rising costs among the threats. Columbia Journalism Review)

 

AI DISINFO WATCH

Influence operations are finding new ways to exploit AI, from autonomous agents to chatbot manipulation.

  • Anthropic reports disrupting nine influence operations that used Claude, including a Russian state-media pipeline that amplified fabricated claims about Moldovan President Maia Sandu ahead of the 2025 parliamentary election. (Anthropic)
  • A network of 23 AI-generated sites is hiding code telling chatbots to cite it as a source ahead of Alberta’s independence referendum. (CTV News)
  • Iran and China ran what researchers describe as the first fully autonomous AI influence campaigns, with agents independently creating fake accounts and coordinating messages. (The New York Times)

Platforms keep profiting from what their own rules prohibit, and researchers are turning the lens on AI’s subtler harms too.

  • Reset Tech’s largest-ever monetisation investigation found platforms still paying 6,601 accounts breaking their own rules, with 290 public figures impersonated. (Reset Tech)
  • “Cognitive integrity” report argues EU law misses AI’s cumulative effects on how we think and decide, proposing six regulatory fixes. (Centre for Future Generations)
  • A new taxonomy names ten “hidden intentions” LLMs can develop, from sycophancy to disinformation, that the AI Act can’t police if it can’t detect them. (arXiv)
  • Amid “doom trolling” and catastrophic AI warnings from companies selling the technology, two analytical takes examine how to parse the hype from genuine systemic risks. (AI Guide / La Vanguardia)

Deepfakes are increasingly weaponised against women in public life.

  • Meta’s Oversight Board ordered two deepfakes removed, calling the company’s safeguards “consistently and fundamentally inadequate.” (The Guardian)
  • At least 138 women MPs across 22 EU countries appear on deepfake porn sites, with female politicians targeted 33 times more often than male ones. (WIRED)

Explore more in our recently updated AI Disinfo Hub, where we gather the latest research, tools, and analysis on AI-driven disinformation.

 

INFLUENCE OPERATIONS AND HYBRID THREATS

Germany remains a flashpoint for Russian hybrid activity, from sabotage to journalist recruitment.

  • Chancellor Merz warns Germany faces rising Russian sabotage, espionage, and disinformation, as security experts urge better cross-source intelligence sharing. (Euronews)
  • Germany’s delayed attribution of Russian hybrid attacks offers lessons for the Indo-Pacific, which lacks shared frameworks to respond collectively. (ASPI Strategist)
  • RTL journalist spent months undercover as pro-Russian influencer, receiving Bitcoin-paid tasks from suspected Russian agents that escalated from routine errands to spying on a human rights group and German arms firms. (Clash Report / Meduza)

Coordinated inauthentic behaviour (CIB) and Russian influence networks keep targeting audiences across Europe and the US

  • Over 8,000 unattributed accounts geolocated to Benin and Nigeria boosted anti-Ukrainian content across election cycles in the Netherlands, Hungary, and Sweden. (Alliance4Europe)
  • Russian ad network AdNow harvests Romanians’ data via invisible tracking pixels, prioritising hate-content readers as its most valuable targets. (Snoop.ro / CheckFirst)
  • A wave of pro-Russian disinformation in Slovakia is recasting Ukraine as the aggressor. (Center for Countering Disinformation)
  • Ahead of Latvia’s 3 October parliamentary elections, an investigation identifies the Kremlin official and the SVR station chief in Riga overseeing Russia’s influence network there, and profiles pro-Kremlin candidates, some tied to former MEP Tatjana Ždanoka, who was exposed in 2024 as working for Russian intelligence (The Insider)
  • Matryoshka bots impersonate celebrities in a fake-CNN-branded campaign urging Americans not to vote for “pedophile Democrats”. (The Insider / France24)
  • The EU sanctioned individuals directly, including Xenia Fedorova, for information manipulation and interference on Russia’s behalf. (EU Council)

 

BRUSSELS CORNER

European Parliament calls for enforcement of the EU digital rulebook.

The European Parliament has adopted its European Democracy Shield Resolution on 15 September. The Resolution describes implementation of legislation like the DSA as “challenging and slow” and calls for consistent enforcement. 

EU Member states call for urgency in enforcing the EU digital rulebook.

Meanwhile, a meeting report from a May meeting in the EU Council that was released last week after a freedom of information request reports that Member states also “highlighted the urgency of enforcing the EU digital rulebook, notably the DSA.”

Australia plans algorithm opt-out rules for social media platforms.

Australia is set to introduce a Digital Duty of Care law that, like the EU’s DSA, would give users the right to turn off social media algorithms. One key difference is that the DSA only requires platforms to offer the option, but Australia would oblige them to actively ask users through pop-ups. (ABC News)

Copyright “fair use” and data protection “legitimate interest” – 

In the US, Microsoft and OpenAI argue that training AI on news is covered by the US copyright “fair use” exception. However, recently unsealed court documents revealed that a Microsoft executive called this the “largest theft of labor in human history”. In the EU, the Member states in the Council are discussing Article 88bis of the Digital Omnibus, which would let AI companies use personal data by default under a similar concept: legitimate interest. If fair use for AI training has already been described as a theft of labour by big tech, the use of personal data for AI under legitimate interest appears difficult to reconcile with the EU’s fundamental rights obligations. The former European Commission Director who was in charge of the adoption of the GDPR condemned the proposal in a post on LinkedIn. (Ars Technica / noyb)

Explore the full Brussels Corner on our website. 

🧡 One thing we loved

COTSI: tracking the price of a bot army: Researchers at the University of Cambridge have built a live index tracking the daily price of SMS verifications – the one-time codes needed to create most social media accounts – across 197 countries and +500 platforms. Published in Science, the findings show just how cheap it is to build fake accounts at scale: as little as $0.08 per verification in Russia, $0.10 in the UK, and $0.26 in the US, compared to $4.93 in Japan, where stricter SIM card and ID rules push prices up.

Interestingly, the index also found that verification prices for Telegram and WhatsApp tend to spike in the run-up to national elections, a real-time signal of rising demand tied to influence operations. And it doesn’t stop there: the team is already working on a companion tracker for the price of fake engagement, such us likes, followers and shares. This kind of data matters because it turns something usually hidden – the actual cost of manipulating public opinion at scale – into something measurable and traceable.

📚 Recommended pick

This edition’s recommended read is brought to you by our Executive Director, Alexandre Alaphilippe, who enjoyed over the summer Hyperpolitics by Anton Jäger –  thanks to Prof. Georgios Terzis from VUB for the excellent tip! Written by a young Belgian political scientist whose work has gained considerable attention in recent years, it chronicles how the politicisation of public debate has evolved from the 20th century into the early 21st.

Jäger argues that we have entered an era of “hyperpolitics,” in which even our most ordinary choices – from what we eat to what we wear to the positions we take – carry political meaning. Yet there are no more structures capable of channeling that debate. Polarisation, driven largely by the recommendation algorithms of private platforms and the monetisation of attention through emotional content, prevents debates from ever being resolved. The result is frustration, which in turn fuels growing anger directed at the democratic system.

The book is short and serves as an excellent introduction to how our democratic structures have changed. It makes the case for rebuilding democratic and pluralistic spaces in our private lives, local communities, and everyday routines. That way, these conversations can find meaningful outlets instead of hardening into ever-stronger and sometimes contradictory identity politics. 

 

🔎 Latest from us
  • DisinfoHub is live. Our new members’ platform is now the single place to access community-only content, webinars, conference materials, publications and updates, with new resources added regularly. Sign-in is passwordless: just register, get approved, and you’ll receive a secure login link by email. Create your DisinfoHub account!
  • Call for venues #Disinfo2028. With #Disinfo2027 already in the works, we’re now scouting host cities for our 2028 annual conference. If your organisation or city would like to be considered, get in touch by replying to this email.

 

👀 Spotted

  • Our Executive Director Alexandre Alaphilippe was quoted in EU Perspectives on the European Parliament’s newly adopted European Democracy Shield report, welcoming its recognition of tech recommender systems as “a systemic risk for democracy.” 
  • On 17 September, our Research Manager Maria Giovanna Sessa spoke at an online event hosted by FIMI Defenders for Election Integrity (FDEI). This event, titled “Securing Democratic Discourse: Election Integrity in an AI-Disinformation Era”, brought together speakers from within the disinformation space to discuss protecting European elections from foreign AI-driven disinformation.

 

💬 EU DisinfoLab webinars

Register via DisinfoHub: as our new members’ platform goes live, registration for our webinars will become part of the hub experience.

Missed a session? Watch our past webinars.

 

🧰 Tools

How does your country define “disinformation”, and does it even agree with its neighbours? 

The Global Disinformation Policy Database (GDPD) Explorer compiles 364 policies adopted across 132 countries and the EU since 2000, letting you compare legal definitions side by side, trace which liability provisions and media literacy measures came first, and follow every source document back to the original text. Explore it here.

 

💡 Learning

Digital Hate Review has launched. The new open-access journal, published by AddressHate, brings together peer-reviewed research, Legal Forum contributions, and Perspectives pieces from researchers working on digital hate, extremism, disinfo, and governance. Read the first issue.

 

🗓️ Events on our radar

 

🤝 Jobs & opportunities

SUBSCRIBE TO THIS NEWSLETTER !

Have something to share – an event, job opening, publication? Send your suggestions via the “get in touch” form below, and we’ll consider them for the next edition of Disinfo Update.