🧵 AI bias isn’t just a chatbot problem. It’s increasingly built into our phones 📱 Our new audit finds the core on-device model powering Apple Intelligence is far from neutral.

🔍 What does TikTok US’s new ownership really mean for users? The transition has raised concerns, from changes to privacy policies to potential censorship. 🗣️Our Head of Research, Salvatore Romano, breaks it down in a new interview with Thomas Bourdeau for @rfi.fr: tinyurl.com/2v8zavmk

🔒Safer Internet Day is a good moment to reflect on how our investigations are helping push the internet in a safer direction — and how we can continue being a catalyst for change in 2026. 🧵

Grok update 🔍 Our latest data shows that safeguards implemented on X did reduce harmful outputs. But on Grok.com, we continue to identify AI-generated images, including some depicting suspected minors.👇

📣 Digital interference in European elections is growing — and it’s not going away. That’s why independent monitoring is so vital. 🧵

Stay tuned! We’ve received many requests in recent days for updates on our Grok report. We’re currently conducting additional monitoring and analysis and will have more to share shortly. 🔍

⚠️ New Investigation: Grok is systematically “undressing” women and generating extremist content. Our latest research shows this isn’t just anecdotal, but a pattern at scale.

Le dernier rapport de l'ONG @aiforensics.org à propos de TikTok montre la profusion de ces comptes qui tentent de surfer sur des "modes" et des contenus outranciers pour bénéficier de la monétisation, et parfois vendre des "formations en IA" aiforensics.org/work/agentic...

Can you still tell when a video is AI-generated? On TikTok, 25% of top search results are now AI-made. We’re releasing a new AI Detection Guide to help journalists, fact-checkers, and everyday users spot AI videos and images. Become your own digital detective. 🕵️ 🔗 tinyurl.com/49z7z9e3

🔍 As the EU calls for “simplification,” our year-long study of LLM chatbots shows a growing regulatory gap. LLMs like ChatGPT and Gemini increasingly act like search engines. But do they fall under the DSA, the AI Act, or somewhere in between?

⚠️ AI chatbots are reshaping how we search. But EU rules haven’t caught up. Our new policy report digs into the risks of LLM-powered search and the gap in Europe’s regulatory approach. So how do we fix it? 👇 Full report + recommendations: tinyurl.com/2p2h77et

TikTok launched a tool to reduce AI content. Our reporting, now covered by @theguardian.com, shows why it's not enough. 🗞️ Agentic AI Accounts push anti-immigrant narratives, sexualized content, and fake news at scale, and TikTok consistently fails to identify and label it. tinyurl.com/yfxca3k3

🚨354 automated AI accounts generated 4.5 billion views on TikTok through more than 43,000 AI-made posts. Our investigation finds Agentic AI Accounts spreading harmful content. Yet TikTok labels around 1.38% of these posts. ⚠️ Full report: tinyurl.com/4kkfm4nn

Academics, civil society and regulators, all coming together to get wonky about the DSA (but more importantly, to hash out ways to help the DSA further efforts to ensure internet integrity in Europe and beyond) 🌟 Thanks again to those who joined our @whattofix.tech / @aiforensics.org workshop!

Billions of people scroll every day. But what’s actually going viral on social media? 🤷‍♀️ We asked the platforms for one simple thing: their most-viewed posts. They all said no 🙃 Help us demand transparency now: mzl.la/3M8Eh9L @algorithmwatch.org @scg-uzh.bsky.social @lessurligneurs.bsky.social

Last month, alongside @mozilla.org @algorithmwatch.org and others, we asked major platforms to share their 1,000 most-viewed posts. Not a single one agreed. If we can’t see what goes viral, we can’t understand how misinformation spreads. Sign the petition for basic transparency. 👇 mzl.la/49HTh8C

Das Internet droht im synthetischen Müll zu versinken. Wikipedia, YouTube, LinkedIn: Überall ist künstlicher Content, auch bekannt als ‘AI Slop’. Wie gefährlich ist das für unsere Wissensinfrastruktur? 👇 www.wissenschaftskommunikation.de/wir-versinke... #Wisskomm @aiforensics.org @kurzgesagt.org

📢 New: TikTok and YouTube failed to address dangerous climate dis/misinformation during Spain’s 2024 floods Our study with @maldita.es.web.brid.gy reveals that such videos remain online a year after the event and appear to have been amplified, receiving over 21 million views. 👇

Europe is rapidly rolling out age-verification — in the name of “child safety.” 🚷 But is it actually working? Our first audit found AgeGO’s selfie feature streamed biometric data straight to Amazon. 🪪 Our new report examines AgeVerif: What we found raises serious concerns. 👇

🚨 From FYP to WW3: Our latest report asks, is TikTok’s algorithm amplifying pro-military narratives? We found: - FYPs prioritized military & war content - Search showed more NATO news & perspectives - Both leaned pro-NATO, incl. praise for Rutte & Macron - “World War III” in ~1 of 25 FYP videos

🚨 "AI Slop" investigation: Platforms failing to label synthetic content We found: - 1 in 4 TikTok search results contain AI imagery - Specialized "Agentic AI Accounts" mass-produce synthetic content - 80%+ of AI content is photorealistic but unlabeled - Instagram desktop users see zero AI labels

📢 In a joint civil society effort, we filed a formal DSA complaint against 𝕏 (formerly Twitter) for violating Article 26(3) — which bans targeting ads based on sensitive data like sexuality, politics, religion, health. 📄 Read the joint statement: aiforensics.org/uploads/Join...

Wanna check if you've been targeted by ads based on sensitive data? Just drop your X portability archive to find out haveibeentargeted.online

Our investigation covered on @pixelsfr.bsky.social by @damienleloup.bsky.social !

How X enables major brands to target your most sensitive personal data - from health conditions to political beliefs - check out our new report! aiforensics.org/work/sensiti...

We also came up with haveibeentargeted.online where you can upload your X data and see whether advertisers have targeted you based on your sensitive characteristics 👇

🧵How X enables major brands to target your most sensitive personal data—from health conditions to political beliefs—all in the name of profit. aiforensics.org/work/sensiti...

💥 Des pubs sur X ciblent sur l'orientation politique, syndicale, religieuse ou les préférences sexuelles. C'est interdit en Europe. X fournit même clé en main des catégories de ciblage potentiellement illégales que ses clients utilisent. Bravo à @aiforensics.org pour l'excellent travail d'analyse !

In @aiforensics.org’s last study authored by @paulbouchaud.com we depict how advertisers use political and other sensitive user data to target ads on X aiforensics.org/work/sensiti...

🧵 We just exposed how TikTok's "Research API" is systematically hiding content from researchers Despite promises of transparency under EU law, the platform is missing 1 in 8 videos from its own research tools: aiforensics.org/work/tk-api

Much focus has been on gaining access to platform data under EU's DSA—but what happens once that access is granted? Democracy Reporting International and @aiforensics.org take a closer look at TikTok’s data access mechanism, exploring how it works in practice. www.techpolicy.press/unpacking-ti...

🧵How TikTok failed Polish democracy during the 2025 Presidential Election Our new investigation exposes serious gaps in TikTok's election content moderation that disproportionately harmed the Polish diaspora : aiforensics.org/work/tiktok-...

📢 NEW INVESTIGATION: Meta approved 46,000+ health scam ads shown 292M+ times to EU users. Meta is systematically failing to moderate fraudulent health ads despite DSA regulations.

"This is a highly adversarial space, we are investing a lot in our detection systems..." Meanwhile, our latest report at @aiforensics.org found Meta reviewed, approved, and displayed fake medicine advertisements over 292 million times to EU users since the DSA took effect.

📊 AI Forensics: A Year of Impact After a pivotal year, we've reached new organizational maturity as leaders in algorithmic auditing & platform accountability. See our annual report to know more: cmsbackend.aiforensics.org/uploads/aif_...

This @bsky.app feature that allows you to choose your browser is refreshing and should be mandatory on every app! In-app browser lock-in is noxious and another way many platforms abuse their dominance

Despite international sanctions, #Meta is accepting payments from Russian propagandists. Our latest report with @aiforensics.org & Reset underscore how Meta has failed to moderate over 8,000 political ads from the Kremlin-linked Social Design Agency (SDA). #SDALeak

Our latest research from @paulbouchaud.com in collab with @checkfirst.network and Reset highlights once again why moderating and regulating Meta and other platforms is necessary

Spoke with @jeremyhsu.bsky.social for this @newscientist.bsky.social piece based on @aiforensics.org's research. newscientist.com/article/2463634-meta-allowed-pornographic-ads-that-break-its-content-moderation-rules/

Free Our Feeds is raising funds to launch a new public interest foundation that puts Bluesky’s underlying technology on a pathway to become an open & healthy social media ecosystem that cannot be controlled by any single company or billionaire (incl Bluesky itself). www.forbes.com/sites/richar...

Social media can be better—for everyone. 🌐 The Free Our Feeds campaign is on a mission to build a new, independent foundation for an ecosystem of interconnected apps & different companies that have people’s interests at heart. Learn how you can help at freeourfeeds.com.

🚨 NEW REPORT: Pay-to-Play: Meta’s Community (Double) Standards on Pornographic Ads Our latest investigation uncovers a glaring double standard in how Meta enforces its content moderation rules. Here's what we found 🧵👇

Blatant double standards on Meta's moderation of sexually explicit content. Read our latest research lead by @paulbouchaud.com, covered today in the Washington Post: aiforensics.org/work/meta-po...

We recently co-signed the Centre for Democracy and Technology’s civil society statement, which calls for enhanced and meaningful transparency from VLOP and VLOSE risk assessments as a mandatory facet of the Digital Services Act.

Our latest investigation shows that Meta's failure to moderate politial ads is exploited by large pro-Russia propaganda network Read the coverage in Politico!