PROMPT ends, the European Narrative Observatory continues

PROMPT closes at Infox sur Seine

On 26-27 February, the Infox sur Seine Festival, held at the University of Paris Panthéon-Assas, brought together a global community of journalists, social scientists, and AI experts to tackle the escalating threats to information integrity.

The event served as the prestigious finale for PROMPT.. As we gathered a global community of journalists, social scientists, and AI experts, the event provided the perfect forum to showcase our final results in "narrative intelligence." Consortium partners unveiled our disinformation detection arsenal, including the latest Wikipedia Sensitivity Barometer. The sessions proved how PROMPT's research has successfully bridged the gap between technical detection and real-world human impact.

Furthermore, Infox sur Seine delved on topics such as the inner workings of Russian disinformation machines to the rise of AI-generated "fake" media brands, connecting the dots between technical detection and human impact. Key sessions explored the psychological factors behind conspiracy theories, the challenges of climate change disinformation, and the "arms race" of generative AI. Blending field ethnography from West Africa and Eastern Europe with cutting-edge fact-checking technologies, the event underscored that safeguarding democracy requires a "whole-of-society" alliance between human expertise and ethical AI.

PROMPT hasn’t been just another brick in the wall. In a fast-paced informational environment—in the midst of our project, Meta dropped its support for European fact-checkers, Musk created Grokipedia, etc.—PROMPT, along with its sister projects, acted as a proof of concept for a more whole-of-society and scaled-up approach to disinformation, which helps consolidate a common grammar to better anticipate and counter disinformation efforts in the European digital space.

→ Le Monde covered our closing event, read here

PROMPT ends but the European Narrative Observatory continues

PROMPT ended on 28 February… but the European Narrative Observatory continues! Under the 3rd Narrative Observatory - project SPINE - will continue to “industrialize” the approach to disinformation. It will combine a strategic data infrastructure with a decentralised, civic-based intelligence community — inspired by models such as Wikipedia. It will tackle 6 interrelated challenges:

- develop scalable and real-time monitoring protocols across 27 Member States and 24 languages.

- provide clean access to data using a multi-pipeline data architecture to avoid overrelying on platforms’ (lack of) cooperation

- build a European Civic Intelligence Community by mobilising volunteers, training regional contributors, and fostering partnerships with Wikimedia chapters and fact-checker networks.

- support the enforcement of the DSA to help transition from voluntary data sharing to structured enforcement, and promote collaboration with open platforms (e.g., Mastodon, Reddit

- support the European Democratic Shield and measures for better data-sharing and interoperability across the EU ecosystem fighting disinformation

- Design open, modular, and upgradable tools to ensure that innovation lives on after the end of the project

Industrialization isn’t about replacing humans—it’s about amplifying their impact. With the 3rd Narrative Observatory, opsci.ai, again consortium leader, hopes to scale efforts to support the European Democratic Shield. Are you interested? Get in touch!

→ For more reflections on what we learnt with PROMPT for the future of the fight against disinformation, read this blog post for the European Journalism Observatory.

Attacks on LGBTQ+, the role of the West in Ukraine, and EU electoral legitimacy: PROMPT Narrative Report 3 is out

Looking at narrative content,coordinated dissemination patterns, persuasion techniques, rhetorical devices and emotional triggers and propagation dynamics, the third PROMPT narrative report examines three targets of disinformation actors:

- the rights of LGBTQI+ community across the France, Romania and Italy, in late 2025-early 2026, uncovering how identity based narratives mobilise moral polarisation, cultural anxiety, and elite resentment

- narrative trends surrounding Russia’s ongoing war in Ukraine in late 2025-early 2026, with particular attention to narrative endurance, strategic reframing and cross platform amplification dynamics.

- the 2024 European Parliament elections, identifying how electoral legitimacy, sovereignty, and geopolitical alignment became focal points of manipulation.

→ Read the report

PROMPT consortium leader opsci.ai joins the EU Code of Conduct on Disinformation

PROMPT developed new tools for the disinformation ecosystem. To scale its work for the better good, consortium leader opsci.ai, an organisation blending pluridisciplinary academic research and practitioner experience, joined the EU Code of Conduct on Disinformation. Opsci.ai has committed to document disinformation prevalence & emergence facilitate the exchange of information, methodologies, evidence, to characterize the prevalence, contents and methods of disinformation, to provide tools for fact-checkers, activists and decision-makers; and more broadly to operationalise a whole-of-society approach into interoperable tools and content.

Where relevant and possible, opsci.ai will share the research developed in the context of the European Narrative Observatory.

The Code of Conduct on Disinformation aims to combat disinformation risks while fully upholding the freedom of speech and enhancing transparency under the Digital Services Act (DSA). As such, the Code will become a relevant benchmark for determining DSA compliance regarding disinformation risks for the providers of VLOPs and VLOSEs that adhere to and comply with its commitments.

→ For more information on the CoC on Disinformation: disinfocode.eu

Wikipedia and the challenges of disinformation: Strategies and recommendations for the resilience of the Wikimedia movement

In an environment of post-truth, Wikipedia finds itself at the intersection of trust and skepticism. While it remains one of the most consulted sources of information globally, it is also a prime target for manipulation by actors seeking to exploit its open-editing model for ideological or political gain - from attempts at "project capture", state-sponsored manipulation by authoritarian regimes and state actors to control narratives on Wikipedia, or manipulations for commercial Influence and reputation management. Recently, AI and information laundering with the rise of AI-generated content, is posing new threats.

Wikipedia’s openness makes it a sanctuary for verifiable knowledge. The platform's transparency, traceability, and community-driven governance provide robust defenses against manipulation. Wikimedia's policy on disinformation is informed by its principles - Neutral Point of View (NPOV), transparency, community governance and proactive moderation, use of reliable sources in particular.

Wikipedia is an active participant against disinformation. It remains a model for digital resilience, in particular thanks to a strong community governance and technological innovation - such as the Wikipedia Sensitivity Meter, developed with Opsci.ai for the PROMPT project. In its recommendations, Wikimedia France suggests several strategic steps: to strengthen community governance; to leverage technology, to reinforce policy advocacy at the EU-level, to advocate for more transparency in AI and enhance collaboration among Wikimedia chapters, in particular in high-risk regions to build resilience against state censorship and political interference.

→ Read the full report here

→ Check out the Wikipedia Sensivity Meter here

PROMPT develops the Wikipedia Treasure Hunt literacy game for disinformation-hunters

Following on 10 beta-testing sessions of its tools, PROMPT introduced a new datathon game, the Wikipedia Treasure Hunt, to familiarize citizens with the risks of information manipulation on Wikipedia, at the Institute of Political Studies at St-Germain-en-Laye.

Using the Wikipedia Sensitivity tool developed by Wikimedia France and Opsci.ai, participants are encouraged to explore potentially vulnerable Wikipedia pages to track known (or unknown!) attempts to manipulate Wikipedia pages. Scrolling through edits, lists of sockpuppets, suspicious sources, is a hands-on way to learn about the vulnerabilities and resilience of Wikipedia, a target of, and shield against, disinformation. Interested in giving it a try? Get in touch!

PROMPT at the EU Media Literacy Expert Group

On 5 March, Opsci.ai and the Erich Brost Institute presented PROMPT’s media literacy activities to the EU Media Literacy Expert Group. In front of 60 media literacy experts representing all EU Member States, we presented our disinformation dissection tools, the PROMPT Massive Online Open Course (MOOC) we developed for future journalists, as well as our recent Guidelines on “AI and journalism”.

The mission of the MLEG is to discover, bring to the light, document and extend good practices in the field of media literacy; facilitate networking between different stakeholders, with the aim of cross-fertilisation; and explore synergies between different EU policies and media literacy initiatives.

Thank you for your interest in our work and for your commitment to information integrity, the cornerstone of our democracies. Let’s keep fighting to defend it!

Best regards from the PROMPT teams