Funded by France 2030, the Confiance.ai technological research program, designed to help industries integrate trustworthy artificial intelligence into their critical systems, concluded in September 2024. Building on the results achieved, the Confiance.ai community has decided to continue its collective investment through a strategy based on three main areas: sustaining the acquired results, deepening research work, and promoting their industrialization at the European level.
Launched in 2021 as part of the Grand Challenge "Securing, Reliability, and Certification of AI-based Systems," the program developed a comprehensive methodology guiding the development and deployment of AI in critical systems. This methodology covers key dimensions such as robustness, explainability, reliability, and regulatory compliance, especially with the AI Act currently being adopted at the European level.

An Open Source Foundation to Structure the Ecosystem

One of the pillars of the new phase of the program is the creation of the European Trustworthy AI Foundation. This foundation aims to:
  • disseminate the tools and methodologies developed under the initial program;
  • federate a European community around shared standards;
  • align technological solutions with regulatory requirements;
  • contribute to the implementation of a European solution for evaluating and labeling AI systems and organizations, within the framework of its involvement in the AI Trust Alliance.

New R&D Initiatives Addressing Emerging Challenges

As AI rapidly evolves, the community identifies new technological challenges to address. These include:
  • generative AI, applied to technical knowledge management;
  • hybrid approaches, combining physical models and machine learning;
  • new forms of human-machine collaboration in complex systems;
  • cybersecurity, which has become a key requirement for any intelligent system.
These efforts will be conducted through concrete use cases, with a focus on transferring to industry, and will enrich the equipped methodology.

A Structured Offering to Support the Industrialization of Responsible AI

To facilitate the adoption of trustworthy AI by companies, several mechanisms are already operational:
  • training offers, including the specialized master's degree "Trustworthy AI" from CentraleSupélec Exed and IRT SystemX;
  • regulatory compliance tools offered by Airbus Protect;
  • industrialization services integrating the constraints of responsible AI, notably provided by Sopra Steria.
A startup from the program will also be created, offering a SaaS solution dedicated to AI component governance, which can be integrated into industrial engineering processes.

Call for Mobilization

Faced with the crucial challenges of AI in industry, the Confiance.ai community calls on all European actors to join its action: actively contribute to shaping the future of industrial AI in Europe.
 Paul Labrogère, Director of IRT SystemX who led this program, concludes:
"Our community is delighted that trustworthy AI has been brought to the forefront during the AI Action Summit. This event highlighted the necessity of mobilizing public, private, and academic actors to build a trust-based ecosystem around AI. Specifically in the field of industrial and responsible AI, Europe can already rely on the largest community worldwide formed by the Confiance.ai program over the past four years. We are already well-positioned for European industries to integrate trustworthy AI solutions to deploy new ultra-competitive innovative services in their markets". 

Translated from Confiance.ai : une feuille de route européenne pour l’industrialisation de l’IA de confiance

To better understand

What is generative AI and how can it be applied to technical knowledge management?

Generative AI refers to AI models capable of creating new data with similar characteristics to the training data. In technical knowledge management, it can generate documents or analyses based on existing databases, thereby simplifying the transfer and updating of knowledge.