Artificial intelligence (AI)
ANSSI supports the development of AI and promotes a risk-based approach to encourage the use of trusted AI systems and to enhance the security of their value chains.
The Agency has been working on the AI question not only to improve the security of AI systems, but also to identify the opportunities and threats these systems may represent with regard to cybersecurity. The development of AI raises a number of cyber-related issues, which may be broken down into three categories:
- The cybersecurity of AI: like all information systems, AI systems are affected by vulnerabilities and may, as a consequence, come under attack. Their security must therefore be ensured. While a number of security measures may be implemented, AI systems are affected by specific vulnerabilities to which security doctrines must be adapted;
- Cybersecurity through AI: the use of AI in cybersecurity provides numerous opportunities, enhancing the effectiveness of security mechanisms and allowing for the automation of certain features;
- Cybersecurity in the face of AI: AI systems also create opportunities for cyber-attackers, particularly with regard to the automation of attacks and to the personalisation and mutation of the threat – developments which further complicate the implementation of effective cybersecurity measures.
ANSSI is actively integrating these cybersecurity challenges into its action plan, operating within the framework of France’s national strategy on artificial intelligence – the latest stage of which was announced in February of 2025 (Opens a new window) (IA : une nouvelle impulsion pour la stratégie nationale)[LGF1] . The objective is to take advantage of the cybersecurity opportunities provided by AI through the development of trusted, secure, and responsible AI which might benefit cyber-defence more than it does cyber-attackers. To that end, ANSSI promotes a risk-based approach to cybersecurity, the implementation of existing cyber regulations, and the elaboration of new rules tailored to the specificities of AI systems.
ANSSI publications on AI
Technical guidance
Building trust in AI through a cyber risk-based approach
7 february 2024
Security recommendations for a generative AI system
27 september 2024
Jointly led international publications
Software bill of materials (SBOM) for artificial intelligence
Publié le Wednesday 13 May 2026
Design Principles for LLM-based Systems with Zero Trust
Publié le Tuesday 12 May 2026
AI Coding Assistants
Publié le Friday 4 October 2024