Information handling: how Inria and VIGINUM are joining forces

What are the respective missions and objectives of VIGINUM and Inria in the fight against information manipulation?

Frédérique Segond, Director of Inria Defence and Security: In accordance with COMP 2024-2028, Inria supports public policy to meet the State's defence and security needs. In the field of combating information manipulation, the institute supports the Ministry of the Armed Forces and the General Secretariat for Defence and National Security on digital issues.

In a geopolitical context marked by the rise of hybrid conflicts, information manipulation – amplified by social networks and generative AI technologies – has become a major strategic lever. It is therefore essential to prevent, detect and analyse these information attacks.

Inria's ambition is to structure a research programme mobilising its project teams, while working closely with the scientific and institutional ecosystem, such as the CNRS, particularly for the humanities and social sciences, and VIGINUM. Inria has also strengthened its expertise by recruiting scientists in the humanities and social sciences who specialise in analysing manipulation mechanisms, at-risk communities and geopolitical contexts.

Anne-Sophie Dhiver, Deputy Head of VIGINUM: VIGINUM's raison d'être is to protect digital public debate affecting the fundamental interests of the nation, and its main mission is to detect and characterise foreign digital interference.

Precisely defined in our founding decrees, foreign digital interference constitutes the digital aspect of information manipulation. It involves, directly or indirectly, a foreign state or a foreign non-state entity, and aims to disseminate, by inauthentic means, allegations or imputations of facts that are manifestly inaccurate or misleading and likely to harm the fundamental interests of the nation.

Why are VIGINUM and Inria joining forces in a three-year partnership? What are the challenges of this collaboration?

Anne-Sophie Dhiver: Faced with the intensification of information manipulation in the digital space, VIGINUM cannot act alone. Building bridges with the academic and scientific community is a strategic priority for the service. Partnering with Inria allows us to combine our expertise in detecting and characterising foreign digital interference with the cutting-edge skills of this leading institution in artificial intelligence, data analysis and information behaviour modelling.

Frédérique Segond: This partnership is a natural fit, as our areas of expertise are highly complementary. VIGINUM deals with specific cases of information manipulation and has access to large volumes of real-time data from its operational activities.

Inria, in turn, brings cutting-edge scientific and technological expertise, particularly in artificial intelligence and data science. This close dialogue enables us to focus our teams' research on key operational issues: characterising information attacks, analysing multimodal content in relation to the humanities and social sciences, and studying dissemination modes and channels, particularly via recommendation algorithms.

The creation of a joint Inria–VIGINUM research laboratory is also being considered in order to structure and strengthen this collaboration in the long term.

How can scientific innovation be translated into operational detection and analysis tools for use by the State?

Anne-Sophie Dhiver: To support analysts in their work detecting and characterising foreign digital interference, VIGINUM has set up a team of data scientists and developers, known as the data lab. The challenge is to be able to implement state-of-the-art methodologies to support investigations. By creating this scientific prize, VIGINUM will be able to discover scientific work that can improve the operational capabilities of the service.

In addition to its operational missions, the data lab is involved in R&D to create open-source tools that enable us to go faster and further in our detection efforts, for example by automatically identifying certain patterns of inauthentic behaviour.

What is the annual VIGINUM-Inria Prize? What are the specific objectives of the Prize and what types of scientific contributions does it seek to encourage?

Anne-Sophie Dhiver: The aim of this scientific prize is to support research in order to accelerate our collective ability to make practical use of innovation, particularly AI, to develop effective operational tools to combat information manipulation. Public authorities and civil society must be able to rely on scientific excellence to better respond to these phenomena and strengthen democratic resilience.

Frédérique Segond: This award aims to support high-level European research by recognising the best scientific work in digital sciences applied to the fight against information manipulation.

The winning contributions must provide new methods, approaches or perspectives that can enrich our joint actions. This initiative is also a means of identifying researchers working on these topics and, where appropriate, inviting them to join our projects – as PhD students or post-doctoral researchers – particularly within the framework of the future Inria–VIGINUM joint laboratory.

Why did you choose to create a European-level science prize, and why with Inria in particular?

Anne-Sophie Dhiver: This prize will recognise and support high-level European scientific work in the field of digital sciences applied to combating information manipulation (LMI). This is a field that is still under development, with many areas yet to be explored. The aim of such an award is to encourage the emergence of ideas and the development of new tools. And what could be more natural than turning to Inria, a leading institution in scientific excellence and innovation, to create this award and promote it? 

How can scientific research help combat the manipulation of information?

Frédérique Segond: Digital sciences are essential for analysing massive volumes of information data. Today, around twenty Inria project teams are working directly on issues related to information manipulation.

A key challenge lies in detection: identifying automatically generated content, spotting coordinated campaigns, and detecting false information disseminated simultaneously across multiple channels. The difficulty is compounded when misleading information is integrated into otherwise truthful content, an issue on which our teams are particularly focused.

Other work focuses on the analysis of textual, audio or video content. By combining these approaches with the expertise of geopolitical specialists, it becomes possible to identify the likely origin of attacks and the communities involved. Finally, studying dissemination methods – particularly the role of recommendation algorithms, influencers and certain public figures – is crucial to understanding the impact and spread of such manipulation.

All of this work contributes significantly to better understanding, detecting and countering information manipulation.