PEACE 2.0: Grounded Explanations and Counter-Speech for Combating Hate Expressions

Greta Damo; Stéphane Petiot; Elena Cabrio; Serena Villata

cs.CLFeb 19, 2026

PEACE 2.0: Grounded Explanations and Counter-Speech for Combating Hate Expressions

Authors: Greta Damo, Stéphane Petiot, Elena Cabrio, Serena Villata

Organizations: Université Côte d’Azur, CNRS, Inria, I3S, France · Université Côte d’Azur, Institut 3IA Côte d’Azur, Techpool, France

Abstract

The increasing volume of hate speech on online platforms poses significant societal challenges. While the Natural Language Processing community has developed effective methods to automatically detect the presence of hate speech, responses to it, called counter-speech, are still an open challenge. We present PEACE 2.0, a novel tool that, besides analysing and explaining why a message is considered hateful or not, also generates a response to it. More specifically, PEACE 2.0 has three main new functionalities: leveraging a Retrieval-Augmented Generation (RAG) pipeline i) to ground HS explanations into evidence and facts, ii) to automatically generate evidence-grounded counter-speech, and iii) exploring the characteristics of counter-speech replies. By integrating these capabilities, PEACE 2.0 enables in-depth analysis and response generation for both explicit and implicit hateful messages.

Figures & tables

Metric	$Exp_{\text{RAG}}$	$Exp_{\text{NoRAG}}$	$Imp_{\text{RAG}}$	$Imp_{\text{NoRAG}}$
Explanations
F	5.00	5.00	5.00	5.00
SO	4.88	4.56	4.80	4.58
I	4.38	2.84	4.64	2.72
SP	4.86	3.78	4.88	4.40
P	4.68	3.52	4.72	3.86

Table 1: Mean human ratings for RAG vs. No-RAG outputs.

Metric	$Exp_{\scriptstyle\text{RAG}}$	$Exp_{\scriptstyle\text{NoRAG}}$	$Imp_{\scriptstyle\text{RAG}}$	$Imp_{\scriptstyle\text{NoRAG}}$
Explanations
Sem. Sim.	0.57	0.49	0.56	0.47
Faithfulness	0.57	-	0.56	-
Perplexity	25.66	37.38	25.43	37.21
Distinct-3	0.98	0.99	0.97	0.99
Hate-Ent.	0.18	0.08	0.12	0.08

Table 2: Automatic metrics results. Abbreviations are for: Semantic Similarity (Sem. Sim.), Entailment (Ent.), Contradiction (Contr.), Evidence (Ev.). Best results are in bold.

Text Generation

Text generation studies methods for producing coherent, useful language from prompts, data, and dialogue context.

See all 422 papers

Counter with Evidence! A Multi-Agent Memory Efficient Reasoning Framework for Hate Category Informed Counterspeech Generation
Aug 24, 2026Sujoy Nath, Aswini Kumar, Tanmoy ChakrabortyText Generation Evidence-Grounded Generation
Indian Institute of Technology Delhi, India
CATCH-ME if you RAG: a dataset of Contextually Annotated multi-Turn Counterspeech against Hate and Misinformation Exchanges
Jun 18, 2026Helena Bonaldi, Genoveffa Martone, Marco GueriniContent Moderation Text Generation
Fondazione Bruno Kessler, Italy · Università Cattolica del Sacro Cuore, Italy
Explain the Flag: Contextualizing Hate Speech Beyond Censorship
Apr 16, 2026Jason Liartis, Eirini Kaldeli, Lambrini Gyftokosta +2Hate Speech Detection
National Technical University of Athens · Datoptron · Independent Researcher +3

Aug 24, 2026J/K move · Enter open · S save

Counter with Evidence! A Multi-Agent Memory Efficient Reasoning Framework for Hate Category Informed Counterspeech Generation

Sujoy Nath, Aswini Kumar, Tanmoy Chakraborty

Indian Institute of Technology Delhi, India

Text Generation Evidence-Grounded Generation Multi-Agent Reasoning

Open paper AskPDF arXiv

Counterspeech effectively neutralizes the impact of online hate. Although prior work explores automated counterspeech generation, it largely emphasizes stylistic control while treating hate speech as homogeneous, overlooking that distinct forms of abuse require fundamentally different counterspeech strategies. To address this gap, we introduce FIRE (Factuality Informed Multi-Agent Reasoning Framework) that first decomposes hate speech into one of the five distinct categories (misinformation, stereotype, conspiracy, dehumanizing, non-factual), and then maps it to a targeted counterspeech style. To facilitate FIRE, we curate FactualCS, a novel dataset of

4,784

instances that provides the annotations regarding hate categories, reasoning traces, and evidence mappings, which are critical elements for grounded generation that are missing in prior work. A comprehensive evaluation across

28

baseline configurations demonstrates that FIRE significantly surpasses existing methods, despite using compact agents (

<

2B). FIRE achieves a

\sim

12 \%

and

\sim

11 \%

improvements in factual and category-specific accuracy respectively, while simultaneously reducing toxicity by

\sim

11 \%

relative to the strongest baselines. Further human evaluation confirms that responses generated by FIRE are significantly preferred over the strongest baselines, underscoring its effectiveness for real-world deployment. These findings show that decomposing the underlying intent of hate speech is essential for generating safe, effective, and contextually precise counterspeech.

Abstract

Text Generation

Explore similar work

Counter with Evidence! A Multi-Agent Memory Efficient Reasoning Framework for Hate Category Informed Counterspeech Generation

CATCH-ME if you RAG: a dataset of Contextually Annotated multi-Turn Counterspeech against Hate and Misinformation Exchanges

Explain the Flag: Contextualizing Hate Speech Beyond Censorship