Violence on TikTok

Following the violence that accompanied the mobilisations of autumn 2026, the French authorities considered temporarily suspending TikTok and other social media platforms. Three arguments are put forward in support of this measure: (i) social media allegedly served to coordinate the violence, (ii) to spread it through images, and (iii) to amplify it artificially, through artificial intelligence or foreign interference operations.

We test these three arguments against nearly one million public videos about the mobilisations, which gathered 16 billion views and 4.5 million comments on TikTok between 1 September and 29 October. Calls to mobilise circulated widely on the platform, but traces of any coordination of violence remain very limited. Images of violence account for 14% of views, and more than half of these come from the media; demands (32% of views) and humour (31%) carry more weight. Comments on videos showing violence condemn these acts nearly ten times more often than they celebrate them, and explicit calls for violence make up 0.27% of comments. Finally, our analyses show neither that violent images enjoy a general visibility advantage likely to encourage their spread, nor that the mobilisations were subject to any major artificial amplification, notably through the use of artificial intelligence or foreign interference operations.

These results do not rule out mechanisms that escape our observation, in particular in private exchanges or personalised recommendations. They nonetheless call into question the relevance of a shutdown that would also hamper uses of the platforms for information, expression and mobilisation.

To take these results further, we call for better researcher access to platform data and invite users to donate their TikTok history.