⚠️ 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.
"The Human Guide to Detecting AI Imagery": For journalists, researchers, and fact-checkers investigating our information ecosystem. And for users defending themselves against a world of AI slop.
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
Quote from TikToks latest Risk Assessment report: TikTok also works with industry groups and civil society to detect undisclosed AIGC. TikTok participates in the Coalition for Content Provenance and Authenticity (‘C2PA”), an open technical standard and content provenance solution that can provide information in a piece of content's metadata about its origins and whether AIGC models were used to create or edit it. This technology helps TikTok to detect AIGC at scale when content with C2PA metadata is uploaded to the Platform. TikTok was the first video-sharing platform to launch the ability to read Content Credentials from C2PA. From Q3 2024 to Q1 2025, 4,781,019 videos on TikTok were auto-labelled with the AIGC tag of “Al-generated” inthe EU.
In their latest risk assessment report (https://www.tiktok.com/transparency/en/dsa-transparency/) #TikTok claims that AI content with the #C2PA label that is uploaded is automatically labeled. I tested this yesterday with one video and it seems this is not true […]
[Original post on chaos.social]
🚨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
I have watched 1,000,000 TikTok videos. Here is my newsletter tiktoktiktoktiktok.substack.com/p/145-the-gr...
📢 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. 👇
Flow chart of the data analysis around using ollama with qwen2.5-vl and other tools like whisper and ffmpeg to analyze TikTok posts.
[TikTok / LLM Analysis]
Our resources are limited and I favor local models over intransparent cloud services. So I build a pipeline around Ollama using a Mac with 32GB RAM and an M2 processor to first process the different channels of the posts […]
[Original post on chaos.social]
⚠️ As TikTok becomes a key space for political engagement, users should know which narratives are amplified—and which might be downplayed.
🔗 Read more: tinyurl.com/rbfjkfmu
🚨 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
Cover of our research "Decoding Secrets With Medals: How Military Insignia Revealed Russia’s Hidden SIGINT Network"
Today, we are proud to publish our latest investigation into the FSB’s 16th Center, publicly accused of operating some of Russia's most sophisticated cyber-espionage groups.
🔗 checkfirst.network/decoding-se...
[TikTok Account Dataset published for BTW2025]
We've just published a list of 2344 #TikTok accounts from political parties and candidates: https://osf.io/6gn8x
The list is a collaborative effort of multiple researchers including University of Potsdam and IDZ Jena as well as independent […]
Our cover depicting "Operation Overload", featuring a blurry silhouette alongside screenshots and laptops.
🔴 NEW REPORT
Alongside @reset.tech, we publish our third investigation into Operation Overload. ⤵️
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 AI Forensics take a closer look at TikTok’s data access mechanism, exploring how it works in practice.
buff.ly/KPZN5do
At AI Forensics we monitored the polish election and found a) #TikTok is still bad at labeling election related content, and b) there were several election fraud conspiracy theories that were not labeled and were seen by millions of people:
https://aiforensics.org/work/tiktok-polish-elections
📊 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_...
screenshot of the TikTok VCE listing "check query task status: Processing" a 100 times
Trying to work with TikToks Virtual Computing Environment (VCE). Requested data on 10(!) Videos.. 20 minutes passed. still waiting.
I helped with this cool project monitoring political content on TikTok before the German Elections. The main findings are explained with the images at text.
