ANTIGONE

ANTIGONE is a European project that approaches the democratic tension between freedom of expression and the fight against disinformation as a deeper question of how disagreement is institutionally and technologically enabled. The project hypothesizes that antagonism in online discourse often stems not from extreme views alone, but from a misalignment between political expression and the institutional or technical conditions through which it is processed. It operationalizes the concept of agonistic democracy, emphasizing the need for digital environments that can mitigate harmful content, but also foster legitimate forms of disagreement. To this end, the project develops theoretical and practical tools to distinguish harmful disinformation from necessary democratic contestation, and to create digital ecosystems that accommodate disagreement without leading to antagonism or symbolic exclusion. ANTIGONE thus pursues four objectives: 1) The project diagnoses how institutional and technological misalignments shape speech and contribute to antagonism. 2) Using human-AI tools and multidisciplinary methods, it examines case studies on “speech grey zones” including climate denial, identity-based violence, and extremism, revealing how harmful narratives drift and mutate across digital environments. 3) ANTIGONE identifies the legal, technical, and governance mechanisms that enable productive disagreement. Through collaboration with media, fact-checkers, and human rights defenders, it explores how formats, legal safeguards, and community tools can support contested speech while protecting vulnerable groups. 4) The project develops frameworks, computational tools, and design principles for agonistic digital ecosystems. Through co-creation workshops and pilots, the project delivers actionable frameworks and guidance, including AI-assisted Open-Source Intelligence (OSINT) tools, GenAI red teaming, and trust-based platform interventions to foster democratic digital environments that support contestation without escalating conflict.

CERTH/ITI contributes to the technical implementation of ANTIGONE by developing AI-assisted methods for analysing information manipulation operations across actors, campaigns and narratives. In particular, CERTH/ITI develops multimodal techniques for detecting hostile actors, identity spoofing, synthetic personas and coordinated inauthentic behaviour, combining behavioural, conversational, visual and cross-platform signals to support human investigators. CERTH/ITI also contributes advanced multimodal narrative discovery and tracking methods capable of analysing text, images and videos to identify emerging narratives, meta-narratives and their evolution across languages, cultures and platforms. Furthermore, CERTH/ITI participates in the design of a red-teaming framework for evaluating the robustness of large language and vision-language models against disinformation, misinformation and harmful content, contributing expertise in AI, multimodal machine learning, OSINT and human-centred AI systems for trustworthy information ecosystems.

Consortium

VRIJE UNIVERSITEIT BRUSSEL, BE
EU DISINFOLAB, BE
QUEEN MARY UNIVERSITY OF LONDON, UK
THE UNIVERSITY OF SHEFFIELD, UK
ETHNIKO KENTRO EREVNAS KAI TECHNOLOGIKIS ANAPTYXIS, EL
EUROPEAN FACT CHECKING STANDARDS NETWORK, FR
AGENCE FRANCE PRESSE, FR
MEDIACONNECT, FR
UNIVERSITEIT VAN AMSTERDAM, NL
INSTITUTE FOR STRATEGIC DIALOGUE GGMBH, DE