Cross-platform research

Automated Moderation Is Here to Stay (2-part series)

Revisits EFF's 2020 position and concludes automated moderation is now permanent across platforms; issues recommendations on transparency and appeals.

Online Domestic Extremism in Canada Data Briefing (Dec 2025-Feb 2026)

Recurring data briefing tracking violent-extremist activity across major platforms in Canada; consistently finds Telegram the leading platform for such activity.

Impact of regulation of children’s digital lives – Phase II

Researchers recorded 108 child-safety/privacy design changes across 70 platforms between 2024 and 2026, but only 5 of the platforms contacted directly responded to information requests. The four largest platforms (Meta, Google, TikTok, Snapchat) have shifted away from protective ‘by default’ design changes toward optional end-user tools such as parental controls, reversing the trend identified in the Phase I report.

Getting Age Assurance Right: A Risk-Based Framework for High-Risk Online Features

Identifies three policy failures jeopardizing children's online safety and proposes a risk-tiered framework for age-appropriate experiences rather than blunt, privacy-invasive age verification.

Artificial Elections 2.0: Generative AI in the 2026 French Elections

AI-generated content in the municipal elections remained largely unlabelled and increasingly normalized, expanding to cartoonish imagery and most prominent on the right/far right.

Rising repression meets global resistance: Internet shutdowns in 2025

Documented 313 shutdowns, a new record, with continued targeted blocking of social media platforms.

DISRUPT White Paper: A Whole-of-Society Approach to Defending Democracy Against Influence Operations

Introduces the DISRUPT toolkit: a three-phase (Prepare, Disrupt, Mitigate) whole-of-society framework for countering influence operations, built on 100+ disrupted cases and input from 50+ European organisations, arguing the gap in Europe’s defences is operational timing, not capability.

Five critical lessons from three years of DSA risk assessments

Reviewing three annual rounds of filings (2023-2025) from Facebook, Instagram, TikTok, X and YouTube, ECNL found reports ‘despite often exceeding 100 pages... still lack the necessary detail’ to understand platforms’ decisions. It identified five recurring gaps: vague risk statements, less diligence on fundamental-rights risks than other systemic risks, unsubstantiated claims, missing detail on stakeholder-consultation feedback, and inconsistent treatment of linguistic/geographic diversity.

'Your report is important to us': unnecessary barriers to submit content reports

Analyzes how platform content-reporting mechanisms impose unnecessary barriers under the DSA.

Dutch Parliamentary Elections 2025 Report

Election-integrity monitoring found coordinated manipulation, new disinformation tactics, a surge in AI-generated content, and clear failures in platform moderation.

Sexualized images on X: what we are doing to stop them and what we expect from the EU

Documented networks of X accounts openly promoting non-consensual 'nudify' tools; a reported post was deemed non-violating, and X refused a DSA Art. 40.12 data request, blocking research (context to the EU's DSA fine on X).

Regulatory challenges & gaps in addressing systemic platform abuse

Election-integrity/FIMI monitoring across four countries reveals regulatory gaps in addressing systemic platform abuse.

War profiteers: online ads and the machinery of propaganda for war

Examines how governments deploy war propaganda through online advertising and argues platforms bear responsibility under international human rights law to prevent monetizing it.

What Kids and Parents Want: Policy Insights for Social Media Safety Features

Research with teens and parents identifies which safety features they actually want and use, arguing policy should center user-desired controls rather than blanket restrictions.

Service-to-Client Interoperability as a Consumer Protection Measure

Argues for mandating 'service-to-client' interoperability, letting independently-developed third-party clients, moderation tools and recommender systems plug into a platform's existing back-end interfaces, as a consumer-protection measure in the EU's forthcoming Digital Fairness Act.

Belarusian Propaganda Watch (April–August 2025)

A monitoring report analysing five Belarusian state-propaganda cases (Independence Day, Slavianski Bazaar, the Zapad-2025 exercises, worker arrivals from Pakistan, and the disappearance of Coordination Council spokesperson Anzhelika Melnikova) across historical, cultural, military, migration and repressive dimensions.

Moldova: Country Election Risk Assessment (CERA)

A pre-election risk assessment rating Moldova’s exposure to FIMI and hybrid threats as "high and rising" ahead of the September 2025 parliamentary elections, outlining three scenarios (limited impact, heightened polarisation, or coordinated hybrid attacks) and urging urgent inter-agency coordination.

Lega Artis’ Rise to Prominence: Inadequate Mechanisms for Curbing Kremlin Propaganda in the Polish Infosphere

Traced how Lega Artis, a 2018 legal blog, became an information-laundering vehicle republishing EU-sanctioned Russian media in Polish, expanding its social reach while circumventing DSA obligations and apparently using AI automation to distribute anti-EU disinformation at implausible volumes.

The Kids Aren't Alright Online: How To Build a Safer, Better Internet

Argues kids' online-safety policy should make platforms safer and more privacy-protective for everyone, focusing on platform design features that facilitate harm rather than blanket restrictions.

Extreme Weather: false claims spread unchecked

Social media companies let false claims about the LA Fires and Hurricane Helene spread unchecked.

Platforms' policies on climate change misinformation (V2)

Updated tracking of how major platforms handle climate disinformation under the DSA, documenting policy rollbacks and weakening.

Content Moderation in the Global South: Four Low-Resource Languages

Synthesizes four case studies showing platforms overwhelmingly invest in English-language moderation, leaving low-resource-language speakers under-protected and subject to both over- and under-enforcement.

Operation Overload: an AI-fuelled escalation ('Overload 2')

The operation escalated using AI-generated content; Telegram is the central hub (673 channels) while ~11,000 X accounts amplify it. TikTok demoted accounts only after 3M+ views.

Moderating Quechua Content on Social Media

Quechua-language users face significantly worse experiences (mis-moderation, reduced reach, lack of language support) than Spanish-language posters.

Moderating Tamil Content on Social Media

Content policies are unevenly defined and enforced for Tamil speakers; automated moderation and Trust & Safety resourcing lag far behind English.

Building the Digital Platform Commission: How To Design a Regulator

Makes the case for a dedicated federal digital regulator and details how to design it to enhance competition, protect consumers, and oversee dominant platforms.

Generative AI and the German Far Right: Narratives, Tactics and Digital Strategies

Documents how German far-right actors use generative AI to produce and spread propaganda across social platforms, with new narrative and digital-strategy tactics.

Emboldened: Internet shutdowns in 2024

Documented 296 shutdowns, the highest since 2016, including social-media-specific blocking during conflicts and elections.

Behind the Curtain: Unveiling Pro-Kremlin Ecosystems in Europe

Maps pro-Kremlin information ecosystems operating across European social media and web infrastructure, detailing coordinated influence and foreign interference tactics.

The 2025 RDR Index: Big Tech Edition

Seventh edition (first full evaluation since 2022): no company scored over 50%; TikTok debuted strongly while X showed RDR's largest-ever single-company decline after cutting its human rights team and transparency reporting.

A Policy Primer for Free Expression and Content Moderation

Lays out a vision for free expression and content moderation given shifts in the media, technological, political and legal landscape, reframing how policymakers should think about platform moderation.

Toxic platforms, broken planet: Online abuse of environmental defenders

Of 204 land/environmental defenders surveyed, 92% had experienced online abuse or harassment tied to their work, with real-world safety and silencing effects.

Moderating Kiswahili Content on Social Media

Kiswahili moderation suffers from limited language investment, sparse local context, and weak automated-system performance, harming users across East/Central Africa.

Rainbow-burning: how social media companies increase risks for LGBTQ+ people in Africa

Documented 214 threatening anti-LGBTQ+ posts across the platforms in five African countries; only 51 were removed when reported, blaming under-resourced moderation for African languages and engagement-driven algorithms.

Doppelganger media impersonation: leveraging domain name dispute resolution

Explores using domain-name dispute resolution as a countermeasure against the Doppelganger media-impersonation operation.

Artificial Elections: Generative AI in 2024 French Elections

Found 51 instances of synthetic imagery, primarily used by right-wing parties to dramatize sensationalist narratives during the 2024 elections; none were labeled as AI-generated.

Joint civil society statement on transparency of risk assessments under the Digital Services Act

The statement, co-signed by ECNL within the DSA Civil Society Coordination Group, sets out what a transparent risk assessment must disclose: detailed methodology, descriptions of algorithmic and advertising systems, risks tied to specific platform functionalities, mitigation measures, which teams/stakeholders were involved, and how audit recommendations fed back into mitigation.

TikTok and Facebook fail to detect harmful disinformation ads (US election)

Weeks before the US presidential election, TikTok approved 50% of disinformation ads and Facebook accepted one; YouTube approved 50% initially but blocked publication pending ID verification.

Hated More: Online Violence Targeting Women of Color Candidates in the 2024 US Election

With University of Pittsburgh: women of color candidates experienced disproportionately high volumes of online abuse, hate, and mis/disinformation compared with other candidates.

Moderating Maghrebi Arabic Content on Social Media

Maghrebi Arabic is frequently mis-moderated, with both wrongful takedowns and failure to catch harmful content, driven by under-investment and ill-suited automated systems.

US Election Platform Preparedness: Comparative Analysis

Assessed platforms across six areas ahead of the 2024 US election; the most alarming common finding was a pervasive lack of transparency about policies, safety efforts and resourcing.

Public Data Access Programs: A First Look

First systematic evaluation of 19 platforms' researcher data-access programs; access varies widely and data quality is poor.

Digital Apartheid in Gaza: Unjust Content Moderation (2-part series)

Documents how platforms, especially Meta (reportedly ~94% compliance with Israeli Cyber Unit takedown requests), disproportionately removed pro-Palestinian content while failing to curb anti-Palestinian hate.

Yet more evidence of Russia's boundless impunity to spread misinformation in the EU

Further documentation of the Doppelganger influence operation's continued reach across social platforms in the EU.

Platforms' AI policy updates in 2024: labelling as the silver bullet?

Reviews platforms' 2024 updates to AI-content policies, arguing reliance on 'labelling' AI content is an insufficient response.

Confusing Google predictions and unbalanced politics on TikTok and Instagram

CrossOver Finland analysis found politically unbalanced content on TikTok and Instagram alongside confusing Google search predictions ahead of the 2024 elections.

Assessing cost-effectiveness: responses to the Doppelganger operation

Assesses responses to the Russian Doppelganger operation that clones media outlets and spreads links/ads across X, Facebook and Telegram.

Ticked off: TikTok approves EU elections disinformation ads in Ireland

TikTok approved all 16 election-disinformation ads ahead of the EU parliamentary elections; X rejected all of them and YouTube rejected all but two.

Shrinking democracy, growing violence: Internet shutdowns in 2023

Annual #KeepItOn documentation of shutdowns and platform blocking worldwide in 2023, tying disruptions to elections, protests and conflict.

Full Disclosure: Stress Testing Tech Platforms' Ad Repositories

With CheckFirst: tested ad-transparency tools of 11 companies; none rated 'ready,' with major data gaps and inaccuracies.

Mozilla commissions CheckFirst to conduct a stress test of Ad Repositories

First comprehensive report on DSA-strengthened platform ad libraries, finding the ad transparency tools failing to meet their intended purpose.

How Automated Content Moderation Works (Even When It Doesn't)

Explainer documenting the automated techniques (hash-matching, classifiers) platforms use to moderate billions of posts, and where these systems routinely fail.

Safe by Default (People vs Big Tech briefing)

Argues engagement-based recommender systems amplifying borderline content is a feature not a bug; calls on regulators to require non-profiling feeds as the default.

Platforms, Promises and Politics: a reality check on election pledges

Finds platforms' election interventions in the Global Majority are template-copied and far weaker than those offered in the US/EU.

In Transparency We Trust?

On platforms' AI-generated-content disclosure mechanisms; finds detection tools unreliable and disclosure approaches inconsistent across platforms.

Reckoning with Mis/Disinformation in 2024 (series)

A reflection series revisiting a decade of mis/disinformation research in the context of the 2024 US election, examining how platform information systems continue to be gamed.

In the Blind Spot: Right-wing Extremists on Alternative and Established Platforms

Analysis of how German-language right-wing extremists build and migrate infrastructure across both mainstream and alternative platforms, exposing moderation 'blind spots.'

Violent and sexualised hate speech targeting women journalists (South Africa)

Nearly all 40 misogynistic test ads (in 4 languages) were approved; Meta and TikTok approved all 40 within 24 hours. Concludes automated moderation is 'not fit for purpose.'

More Tools, More Control: Young Users on Handling Unwanted Messages

Young users want more granular controls to manage unwanted messages; current platform tools are insufficient, with design recommendations for youth safety that preserve autonomy.

Platforms' policies on AI-manipulated and generated misinformation

Maps how major platforms address AI-manipulated and AI-generated misinformation in their policies.

Platforms' policies on climate change misinformation (V1)

Maps how the five platforms address climate misinformation, revealing uneven and weak climate-specific rules.

Towards Meaningful Fundamental Rights Impact Assessments under the DSA

The paper argues DSA Article 34(1)(b) fundamental rights impact assessments (FRIAs) risk becoming ‘tick-box’ exercises without harmonized minimum standards. It sets six benchmarks papers must meet — governance grounded in the EU Charter, full scope, sufficient methodological detail, public transparency, a harmonized-but-flexible structure, and regular external stakeholder consultation — plus separate benchmarks for algorithmic content-moderation risks.

A world of online hate and lies: Mapping our investigations

A synthesis of 10+ country investigations concluding platforms systematically fail to enforce their own hate-speech/disinformation ad policies, with weaker moderation outside the US.

Fixing Recommender Systems: From Identification of Risk Factors to Meaningful Transparency and Mitigation

Sets out six hypotheses on how recommender-system design choices (engagement-driven ranking, addictive features, data-driven personalisation) generate DSA Article 34 'systemic risks,' plus a technical checklist of disclosures VLOPs/VLOSEs should make to regulators, auditors and researchers.

Platforms' policies on elections misinformation

Maps and compares how five major platforms address election misinformation in their policies.

Platforms' policies on health misinformation

Maps and compares how five major platforms address health misinformation in their policies.

'We need to kill them': Xenophobic hate speech approved by Facebook, TikTok and YouTube (South Africa)

Violent xenophobic test ads were approved by all three platforms; nearly every ad passed, with only one Facebook rejection.

Lost in Translation: Large Language Models in Non-English Content Analysis

Multilingual LLMs used for content moderation perform substantially worse in non-English (especially low-resource) languages; cautions against treating them as a fix for global moderation.

State Hate: How Iran's Press TV Uses Social Media to Promote Anti-Jewish Hatred

With ADL: Iran's state broadcaster weaponizes social platforms to spread antisemitic conspiracy content.

Extreme and violent anti-LGBTQ+ hate approved for publication by leading social media platforms

Test ads with extreme anti-LGBTQ+ hate were almost all approved: YouTube and TikTok approved all 10 ads, Facebook rejected only two, and all three approved the 'burn all gays' ad.