News
breadcrumb arrow
Political cyberbullying: Nearly one in two comments is problematic on X
Political cyberbullying: Nearly one in two comments is problematic on X
Lyess Meddahi
July 22, 2026

SafeBear publishes the new edition of its monthly cyberbullying barometer targeting the main French political figures on the social network X. Conducted on posts published between 1 and 31 May 2026, this study analyses nearly 230,000 comments received by forty French political figures using SafeBear's proprietary technology for detecting and classifying toxic content. Key finding: 43% of the comments analysed are problematic, confirming the high level of verbal violence observed on the platform.

A structural toxicity in online debate

  • 229,730 comments were analysed in May 2026, compared to 218,686 in April, representing a 5% increase.
  • 98,030 comments were identified as problematic, representing a stable overall toxicity rate of 43%.

Political figures heavily exposed

  • Jean-Luc Mélenchon remains the political figure receiving the highest number of toxic comments (16,118), ahead of Emmanuel Macron (7,763) and Gabriel Attal (7,564).
  • Jordan Bardella ranks fourth with 7,255 problematic comments.
  • The most active figures on X during May are Jean-Luc Mélenchon (145 posts), Marine Le Pen (129), Jordan Bardella (128), Éric Ciotti (123), Nicolas Dupont-Aignan (115), Florian Philippot (112) and François Ruffin (103).

Variable intensity depending on profile

  • As a proportion of comments received, Nicolas Sarkozy records the highest toxicity rate (64%), ahead of François Hollande (57%) and Jean-Luc Mélenchon (54%).
  • The largest increases in toxic comments compared to April concern Florian Philippot (+174%), Marine Le Pen (+82%), Jean-Luc Mélenchon (+76%) and Gabriel Attal (+58%).
  • Conversely, Aurore Bergé records the largest drop in problematic comment volume (-65%).

Well-identified forms of violence

Insults represent the most frequent form of violence observed on X. Of the 98,030 problematic comments detected in May, SafeBear recorded 60,528 insults, 23,302 denigration messages, 17,599 pieces of identity-based hate content and 4,686 messages containing threats.

"When nearly one in two comments addressed to a public figure contains a form of verbal aggression, we are no longer dealing with marginal behaviour but with a structural phenomenon that affects the quality of democratic debate," says Lyess MEDDAHI, Strategy Director at SafeBear.

CYBERBULLYING OBSERVATORY ON THE SOCIAL NETWORK X
PUBLIC FIGURES TWEETS
POSTED
COMMENTS RECEIVED PROBLEMATIC
COMMENTS
TOXICITY RATE*
MAY 26 MAY 26 APR. 26 MAY 26 APR. 26 MAY 26 APR. 26
Jean-Luc Mélenchon 145 30 084 16 166 16 118 9 165 54% 57%
Emmanuel Macron 77 21 099 24 649 7 763 9 955 37% 40%
Gabriel Attal 90 17 270 10 461 7 564 4 791 44% 46%
Jordan Bardella 128 17 212 17 016 7 255 7 411 42% 44%
Marine Le Pen 129 15 874 9 278 6 331 3 483 40% 38%
Florian Philippot 112 11 987 4 315 5 198 1 894 43% 44%
Olivier Faure 61 11 394 13 766 4 924 6 235 43% 45%
Bruno Retailleau 71 11 030 10 346 4 937 4 330 45% 42%
Sarah Knafo 61 10 171 10 045 3 450 2 889 34% 29%
David Lisnard 84 7 851 6 699 1 935 2 084 25% 31%
Eric Zemmour 53 7 702 6 404 3 351 2 984 44% 47%
Yaël Braun-Pivet 37 6 535 6 726 2 981 3 175 46% 47%
François Ruffin 103 6 209 5 253 2 497 2 044 40% 39%
N. Dupont-Aignan 115 5 604 7 164 2 361 3 035 42% 42%
Marion Maréchal 55 5 473 5 655 2 632 2 673 48% 47%
Clémentine Autain 67 4 220 4 379 1 828 1 978 43% 45%
Ségolène Royal 36 3 857 4 781 2 012 2 436 52% 51%
Eric Ciotti 123 3 575 4 724 1 286 1 305 36% 28%
Gérald Darmanin 69 3 269 2 996 1 238 1 062 38% 35%
Dominique de Villepin 33 3 214 3 862 1 352 1 925 42% 50%
Nathalie Arthaud 65 2 874 4 784 1 160 2 520 40% 53%
Élisabeth Borne 31 2 740 450 1 076 167 39% 37%
Fabien Roussel 75 2 725 4 101 1 024 1 822 38% 44%
Laurent Wauquiez 20 2 212 2 720 899 1 177 41% 43%
Aurore Bergé 36 2 057 5 067 933 2 693 45% 53%
Jérôme Guedj 31 1 812 2 826 821 1 395 45% 49%
Thierry Breton 13 1 672 659 858 284 51% 43%
Valérie Pécresse 21 1 559 1 699 588 630 38% 37%
Edouard Philippe 5 1 528 459 744 236 49% 51%
Raphaël Glucksmann 7 1 426 1 047 591 500 41% 48%
Xavier Bertrand 26 1 357 1 377 677 660 50% 48%
Bernard Cazeneuve 13 965 508 373 182 39% 36%
Carole Delga 26 856 1 487 304 578 36% 39%
François Asselineau 17 607 8 158 202 3 296 33% 40%
Rachida Dati 13 566 186 233 72 41% 39%
Marine Tondelier 78 444 4 135 202 1 988 45% 48%
Michel Barnier 17 188 1 992 67 747 36% 38%
François Hollande 6 153 1 524 87 738 57% 48%
Nicolas Sarkozy 1 146 94 64%
Boris Vallaud 43 113 130 42 46 37% 35%
Gérard Larcher 8 100 692 42 301 42% 43%
TOTAL 2 201 229 730 218 686 98 030 94 886 43% 43%

Source: SafeBear, May 2026
*Toxicity rate: ratio between the number of comments received and the number of problematic comments


CYBERBULLYING OBSERVATORY ON THE SOCIAL NETWORK X
Categories of toxic messages – May 2026
TOTAL OF WHICH
PUBLIC FIGURES Problematic
comments
Disparagement Obscene Insults Identity-based
hate
Threats
Jean-Luc Mélenchon 16 118 3 730 59 10 207 3 762 782
Emmanuel Macron 7 763 1 702 61 4 398 1 703 461
Gabriel Attal 7 564 1 782 49 5 337 744 230
Jordan Bardella 7 255 1 816 25 4 726 1 005 311
Marine Le Pen 6 331 1 941 21 3 550 1 130 344
Florian Philippot 5 198 1 177 37 3 004 454 376
Olivier Faure 4 924 1 115 17 3 482 676 141
Bruno Retailleau 4 937 1 192 14 2 980 909 232
Sarah Knafo 3 450 750 31 1 832 937 204
David Lisnard 1 935 407 3 1 102 354 128
Eric Zemmour 3 351 645 9 1 712 1 251 143
Yaël Braun-Pivet 2 981 845 19 1 627 477 134
François Ruffin 2 497 546 8 1 718 332 100
N. Dupont-Aignan 2 361 474 15 1 515 208 139
Marion Maréchal 2 632 671 19 1 317 875 100
Clémentine Autain 1 828 361 14 1 256 304 80
Ségolène Royal 2 012 499 18 1 275 383 54
Eric Ciotti 1 286 297 5 762 216 94
Gérald Darmanin 1 238 356 22 612 262 95
Dominique de Villepin 1 352 353 6 818 198 53
Nathalie Arthaud 1 160 204 3 811 146 65
Élisabeth Borne 1 076 259 5 739 111 37
Fabien Roussel 1 024 212 6 667 133 39
Laurent Wauquiez 899 225 3 669 48 38
Aurore Bergé 933 262 12 474 230 38
Jérôme Guedj 821 171 3 565 170 30
Thierry Breton 858 228 1 613 63 34
Valérie Pécresse 588 146 4 384 88 24
Edouard Philippe 744 180 1 477 75 55
Raphaël Glucksmann 591 145 1 401 54 18
Xavier Bertrand 677 147 2 485 104 20
Bernard Cazeneuve 373 105 2 251 32 20
Carole Delga 304 84 1 198 40 18
François Asselineau 202 56 2 90 32 7
Rachida Dati 233 71 1 119 46 13
Marine Tondelier 202 51 0 145 21 5
Michel Barnier 67 14 1 46 8 0
François Hollande 87 20 0 61 2 11
Nicolas Sarkozy 94 30 1 56 4 11
Boris Vallaud 42 13 0 23 7 1
Gérard Larcher 42 20 0 24 5 1
TOTAL 98 030 23 302 501 60 528 17 599 4 686

Source: SafeBear, May 2026
A comment may fall into multiple categories.


Definition of indicators Problematic messages are classified into 5 categories:

  • Disparagement: personal attacks & defamation. Rude or disrespectful messages intended to belittle, offend or embarrass, and to discredit;
  • Obscene: sexual cyberviolence. Messages with offensive language, sexual content or explicit material unsuitable for the general public;
  • Identity-based hate: racist, antisemitic, Islamophobic, sexist, homophobic, xenophobic hatred. Messages that encourage hatred or discrimination based on identity (race, religion, gender, etc.);
  • Insults: messages containing insults or disparaging remarks targeting individuals or groups;
  • Threats: messages containing threats of violence or harm towards individuals or groups.
Other articles
How to React to Social Media Harassment?
How to React to Social Media Harassment?
Regardless of your age, gender, beliefs or origins, cyberbullying can affect any
Nicolas
July 2, 2026
How to Talk About Cyberbullying with Your Child?
How to Talk About Cyberbullying with Your Child?
Cyberbullying is a serious problem that affects many children. It's important to
Nicolas
July 2, 2026
Cyberbullying: Definition and Regulations
Cyberbullying: Definition and Regulations
What is cyberbullying and what are the consequences if you are found guilty?
Nicolas
July 2, 2026
French innovation that secures online exchanges for everyone
Follow Us
facebook x
linkedin instagram
© 2026 SafeBear