Capgemini Leads EU Team on AI Cyber Risk
The European Commission's AI Office has chosen Capgemini to lead a consortium assessing cyber offence risks posed by advanced general-purpose AI models.
Updated on Nov. 5, 2026 in Artificial Intelligence
Live Poll
Do you trust government regulation to keep powerful AI systems safe from cyber misuse?
The European Commission's AI Office selected Capgemini to lead a consortium supporting cyber threat modeling and risk assessment of advanced AI systems. The assignment forms part of the EU's AI Safety program and is under Lot 2 of procurement procedure EC-CNECT-2025-OP-0032.
Why it matters
Advanced general-purpose AI models are gaining autonomous and cyber-relevant capabilities such as automated vulnerability discovery and exploit generation. The consortium will help the AI Office identify systemic risks and enforce the AI Act's obligations for GPAI providers.
The consortium will model emerging cyber offence risks, develop evaluation tools, design technical compliance methodologies, and conduct on-demand technical assessments of General-Purpose AI models. Capgemini's team spans more than 410,000 members across over 50 countries.
The players
Capgemini
A global consulting and technology services company headquartered in Paris with nearly 60 years of expertise.
European Commission AI Office
The office within DG CNECT responsible for overseeing general-purpose AI and enforcing the EU AI Act.
Andrea Falleni
CEO of Capgemini's Southern Europe region and a member of the company's Group Executive Board.
The details
Capgemini leads a consortium that includes Wavestone and the French public institution LNE (Laboratoire National de Métrologie et d'Essais). The group brings together expertise in cybersecurity risk management, operational security, AI evaluation and conformity assessment.
Timeline
November 5, 2026: Capgemini's selection was announced in Paris.
Tech Race
The assignment operationalizes the EU AI Act's obligations for general-purpose AI providers by building the technical assessment capacity the AI Office needs to enforce them. It positions European regulators to evaluate cyber offence risks from frontier models rather than relying solely on developer self-reporting.
The assessments are intended to inform how General-Purpose AI models are evaluated for cyber offence risks before and during deployment. Users of these models may eventually see stronger safeguards tied to the EU's compliance requirements.
The takeaway
Regulators are building independent technical capacity to test frontier AI systems for cyber misuse rather than depending only on what developers disclose. Readers following AI policy should watch how these evaluation methods are applied under the AI Act.
Further reading
For broader context on how regulators are approaching these systems, see our coverage in Artificial Intelligence.
Source note: This article includes information reported by The Manila times.
Live Poll
Do you trust government regulation to keep powerful AI systems safe from cyber misuse?
