Scientific research · 3 November 2026

Did TikTok fuel the violence of autumn 2026 in France?

In autumn 2026, France saw two social movements: a high-school student movement, with school blockades on 1 October, then a general mobilisation on 17 October, with demonstrations marked by major clashes. The authorities then considered temporarily suspending TikTok. We analysed nearly one million public videos.

See the results ↓

The study

Nearly one million videos, from 1 September to 29 October

Several hundred high schools blockaded on 1 and 2 October, between 550,000 and 800,000 demonstrators on 17 October. We studied the videos posted from France about these mobilisations. Their audience follows these two waves.

  • 964,000 videos
  • 16 billion views
  • 4.5 million comments
Read the full note (PDF, 30 pages)

Findings

Three arguments against social media, tested on TikTok's public videos

Three arguments were put forward to justify restricting social media. Here is what we found.

1 Was TikTok used to coordinate the violence?

Calls to mobilise circulated widely. Calls for violence, hardly at all.

Locations mentioned in advance, referrals to private messaging apps, identical messages posted simultaneously: such traces are rare.

Fewer than 0.03% of comments refer to a private messaging group.

Cumulative views before 17 October

Depending on the day, videos calling for violence gather at most 0.11% of views.

2 Did images of violence dominate?

Images of violence account for 14% of views, less than demands or humour.

A video can fall into several categories, so the shares do not sum to 100%.

Share of views of videos about the mobilisations

The visibility of violence is due first and foremost to the media.

“Small creators” have fewer than 50,000 followers; “others” are influencers, anonymous pages and other accounts.

Small creators Media Others

Videos posted

Media: 1.6%

All views

Views of violent videos

Under videos showing violence, comments condemn these acts far more often than they celebrate them.

Explicit calls for violence make up 0.27% of all comments.

Comments under videos showing violence

3 Was the violence artificially amplified?

Showing violence is associated with fewer views for small creators, but more for the media.

Content labelled as AI accounts for less than 3% of views. None of our six indicators reveals any large-scale interference operation.

Views of a video showing violence, compared with the same account's videos without violence, day of publication taken into account

What these results do not say. They cover public videos only. They do not rule out what happens in private exchanges or personalised recommendations. That is why we need your watch history.

First page of the note

The note

Violence on TikTok

What do 16 billion views reveal about the mobilisations of autumn 2026?

Read the note (PDF) Version française (PDF)

To cite: Benjamin Guinaudeau, Isabelle Guinaudeau, Hugo Subtil, « Violences sur TikTok », Observatoire du bien-être du Cepremap, n°2026-21, 03/11/2026. This version is an English translation of the original note published in French (OBE Note 2026-21, “Violences sur TikTok”). In case of discrepancy, the French version prevails.

About the authors

  • Benjamin Guinaudeau

    Professor of political science

    Université Laval, affiliated with New York University (CSMaP)

  • Isabelle Guinaudeau

    CNRS researcher

    Centre Marc Bloch and Sciences Po (CEE)

  • Hugo Subtil

    Postdoctoral researcher

    University of Zurich