Propaganda Forensics: Recovering the Generation Pipeline of an AI-Driven Influence Campaign

Benjamin Icard; Elouan Vuichard; Louis Lefebvre; Lila Sainero; Thomas Girault; Alice Breton; Tanguy Launay; Gauvain Bourgne

cs.AIAug 16, 2026

Propaganda Forensics: Recovering the Generation Pipeline of an AI-Driven Influence Campaign

Authors: Benjamin Icard, Elouan Vuichard, Louis Lefebvre, Lila Sainero, Thomas Girault, Alice Breton, Tanguy Launay, Gauvain Bourgne, +7 more

Organizations: LIP6, Sorbonne University, CNRS · IRL Crossing, CNRS · AIRBUS · SIPA Ouest-France · IRISA, CNRS

Abstract

We present a forensic analysis of the generation pipeline behind a recent AI-driven influence campaign. We introduce PROPAGIA, a corpus of 2,646 propagandist French articles from the Storm-1516/CopyCop campaign disclosed by VIGINUM and INSIKT GROUP in 2025. For comparison, we rely on SIPA, a corpus of human-written French mainstream press from the same period. Using topic modeling, vagueness and sentiment analysis, we first isolate persuasion techniques characteristic of propaganda, with PROPAGIA far exceeding SIPA in vagueness, subjectivity and negativity, and citing fewer sources. We then find prompt instruction leaks on 50 of the 84 PROPAGIA websites, including a verbatim ten-point editorial specification accounting for several of these differences, together with high cross-article redundancy. Finally, we show that rewriting-based detection supports INSIKT GROUP's attribution to the Llama 3 family, but also suggests the involvement of Mistral-family models.

Figures & tables

Figure 1: Comparison of coverage between SIPA and PROPAGIA for the top 20 topics

Panel of Figure 2 — Figure 2: Comparison of the SIPA and PROPAGIA corpora in terms of VAGO scores (left), and quotation scores (right).

Figure 3: Mean objectivity scores by topic and source; error bars show the standard error of the mean (SEM).

Figure 4: Sentiment progression of SIPA and PROPAGIA articles using the TabularisAI model.

Panel of Figure 5 — Figure 5: Prompt instruction checklist found verbatim in a PROPAGIA article assigned to Topic 8.

Figure 6: Distribution of articles sharing at least $x\%$ of their sentences with at least another within each corpus. Median number of sentences per article: 18 for SIPA , 14 for PROPAGIA .

Instruction	Indicator	SIPA	PROPAGIA
1, 3	Articles with $>50\%$ sentence overlap	2.5%	10%
4	Quotation score	.32	.11
5--7	Subjectivity	.48	.54
	Objectivity	.74	.55
8	Negativeness trend	Lower, stable	Higher, rising
9	Putin-specific sentiment	Not evaluated

Table 1: Leaked instructions and corpus-level indicators.

Figure 7: Mean fuzzy score (original vs. rewrite) by corpus and rewriting model, averaged across seven prompts. Black diamonds indicate means. Higher values indicate less editing by the rewriting LLM.

Figure 8: Distribution of the number of articles per impersonating website in PROPAGIA .

Appendix figures & tables6 assets

Supplementary material from the paper’s appendix.

Appendix

Statistic	Value
Number of articles	2,646
Number of impersonating websites	84
Articles per website (mean)	31.5
Articles per website (median)	25
Tokens per article (mean)	250.5
Tokens per article (minimum)	100

Appendix

Table 2: Corpus statistics for PROPAGIA . The token minimum and maximum are the observed range of the collected articles, not an inclusion criterion.

Figure 9: Comparison of emotion and negativeness scores between the SIPA and PROPAGIA corpora, based on the French FEEL emotion lexicon. Diamonds show the mean of each distribution.

Figure 10: Sentiment progression of SIPA and PROPAGIA articles using the zero-shot Qwen3.6-35B-A3B model.

Figure 11: Sentiment progression of SIPA and PROPAGIA articles using the zero-shot mDeBERTa-v3 model.

Rewriting model	GPU hours
Dolphin	148.7
Llama	138.0
Mistral	115.7
Gemma	83.0
Zephyr	71.0
Qwen	59.0

Appendix

Table 3: Per-model compute cost of the RAIDAR rewriting analysis.

Panel of Figure 13 — Figure 13: Full-length article from PROPAGIA , translated from French into English and reproduced verbatim, assigned to Topic 8 ( “Epstein Scandal: Documents, Power Networks, and Political Fallout” ).

Explore similar work

IO Factory: Simulating AI-Enabled Influence Campaigns at Scale
Aug 11, 2026Lukasz Olejnik, Wenchao Dong, Jonas R. Kunst +4Agent-Based Modeling Automated Red Teaming
Department of War Studies, King’s College London, London, United Kingdom · Independent researcher · Max Planck Institute for Security and Privacy, Bochum, Germany +4
InfoOps Bench: A live information operations safety benchmark
Jul 30, 2026Dorian Quelle, Lisa-Maria Neudert, Jonathan Bright +1LLM Safety Benchmarks Language Model Safety Evaluation
SlopShape: Identifying AI-Generated Commercial Web Content
Sep 14, 2026Jochen MadlerAI-Generated Text Detection Language Model Fingerprinting
Sitefire

Aug 11, 2026, cs.AIJ/K move · Enter open · S save

IO Factory: Simulating AI-Enabled Influence Campaigns at Scale

Lukasz Olejnik, Wenchao Dong, Jonas R. Kunst, Signe Riemer-Sørensen+3

Department of War Studies, King’s College London, London, United Kingdom · Independent researcher · Max Planck Institute for Security and Privacy, Bochum, Germany+4

Agent-Based Modeling Automated Red Teaming

Open paper AskPDF arXiv

We introduce IO Factory, an AI-driven framework for simulating information and influence campaigns as fully integrated, traceable processes. The threat of digital manipulation now extends beyond persuasive text from individual language models to AI swarms, i.e., persistent groups of coordinated agents that adapt to platform feedback and disguise organized campaigns as ordinary social interaction. Because such campaigns cannot be identified from isolated messages alone, they must be analyzed across a continuous spectrum of planning, platform action, exposure, interpretation, measurement, and adaptation. IO Factory represents this process inside a controlled simulated platform, linking actor roles, platform actions, exposure records, structured model-based evaluations, and configured changes in the simulated population. We implement the architecture and evaluate it across configurations of up to 100,000 agents. The results show that IO Factory executes campaign timelines at scale and produces inspectable evidence of exposure and measured movement in configured belief variables. By recording the actors, objectives, action constraints, exposure paths, and measurement rules used in each run, IO Factory supports reproducible research and red-team analysis of coordinated influence.

Abstract

Explore similar work

IO Factory: Simulating AI-Enabled Influence Campaigns at Scale

InfoOps Bench: A live information operations safety benchmark

SlopShape: Identifying AI-Generated Commercial Web Content