Skip to content
France
File

Orange Cyberdefense and Mistral: AI for security operations

This dossier follows the announced integration of Mistral models into Orange Cyberdefense services, intended uses and supervision requirements.

Last updated : 1 key moments

Where things stand

As of 30 October 2026, Orange Cyberdefense announces early access to a cybersecurity variant of Large 4. Integration is to be gradual; no quantified operational benefit is established.

Orange Cyberdefense to test a cyber variant of Mistral Large 4

Orange Cyberdefense announced on 29 October 2026 that it had joined Mistral’s design partner programme. The agreement provides early access to a Large 4 variant intended for advanced cybersecurity operations.

Mistral AI logo
Logo of Mistral AI, Orange Cyberdefense’s announced technology partner. — Mistral AI / Pantarch, Wikimedia Commons — public domain (PD-shape).

Initial work will focus on penetration testing, technical audits and post-attack analysis. The Orange subsidiary plans to integrate the tools gradually into selected customer services, alongside analysts.

According to the statement, experts will retain responsibility for critical actions, with access controls, traceability and model security testing. Computing is to take place in Europe. Mistral says access to the specialised variant requires vetting and contractual safeguards, with testing limited to systems owned or explicitly authorised.

This is an integration and evaluation programme: the sources provide neither a detailed timetable for wider deployment nor a quantified reduction in response times.

Sources 3

Orange Cyberdefense and Mistral: AI for security operations · Moojaz