Summary
Our research had two goals: to document the current landscape of disinformation on social media, and to test how well the Trusted Flagger mechanism works in practice under the DSA. Over two weeks, we screened thousands of videos and identified 72 items as potentially unlawful or problematic, falling into five categories: AI-generated videos, general disinformation, violent or attack-glorifying content, neo-Nazi or otherwise unlawful extremist content, and health-related misinformation.
Problematic content was particularly easy to encounter on TikTok, where interacting with a single such video quickly triggered recommendations for more similar content in the ‘For You’ feed - pointing to a broader risk: recommender systems can actively amplify harmful material rather than contain it.
The results also show major weaknesses in platform moderation:
Moderation showed clear weaknesses.
- Of 21 AI-generated videos reported, 16 were classified as "no violation," despite many being designed to manipulate public perception around war and geopolitical conflict.
- Of 24 videos reported for glorifying attacks, 9 were also dismissed as "no violation” - including content portraying perpetrators as celebrities or role models.
- TikTok often failed to act consistently even in cases involving Nazi symbolism, glorification of National Socialism, or antisemitic content.
The Trusted Flagger mechanism outperformed private reporting, but not reliably: while it led to removals more often, response times were inconsistent, and in one case a private report was processed faster than a Trusted Flagger report for similar content. Private reporting was largely ineffective, with over 81% of reports dismissed as "no violation." On Instagram, none of the seven items reported as Trusted Flagger were removed.
Recommendations
Ensure consistent Trusted Flagger response and removal: TikTok and Instagram should apply consistent standards to Trusted Flagger reports.
Reduce algorithmic amplification of flagged content categories: TikTok should adjust its recommender system so that engagement with disinformation or extremist content does not lead to increased recommendation of similar material in the ‘For You’ feed.
Guarantee functional Trusted Flagger status on Instagram: Meta should ensure that Trusted Flagger reports on Instagram result in effective review and action; in this study, none of the reports submitted through this channel led to removal.
TGuard — funded by Kiras Sicherheitsforschung, FFG, and Bundesministerium Finanzen.
This research was conducted as part of the "TGuard" project and funded by the KIRAS security research programme of the Federal Ministry of Finance. For more information about the project visit: https://www.tguard.at/
