Cyber & IT Supervisory Forum - Additional Resources

A MULTILAYER FRAMEWORK FOR GOOD CYBERSECURITY PRACTICES FOR AI June 2023

ANNEX I: QUESTIONNAIRE

Goal of the questionnaire: The proposal for a regulation (AI Act) provides various harmonised rules. The goal of the questionnaire is to assess the level of preparedness of authorities for monitoring and enforcement of these future requirements by the NCAs and evaluate whether certain practices/requirements established under the AI Act have been already implemented at the level of the MS. Methodology : Our questionnaire is based on the cybersecurity concepts described in the FAICP framework, on the main principles (related to cybersecurity) of the proposed AI Act and the coordinated plan on AI and on the harmonised rules related to cybersecurity as reported in the explanatory memorandum (see Table 2). Table 2: Summary of AI cybersecurity aspects relating with policy areas and national strategies

Policy areas

Public and private sector initiatives

Cybersecurity of AI

Human capital

Education/training / lifelong learning Labour market Intelligence/skills Demand

Enhancement of AI-related skills Education reforms Initiatives targeting teachers and educators Lifelong learning and continuing education: Upskilling and reskilling Future labour market: skills intelligence

Inclusion of AI security in all forms of education/training Awareness initiatives about security of AI

Hands-on skills and practical capabilities of AI cybersecurity

From the lab to the market

R & D Innovation

Support AI research Initiatives for innovations towards business growth and increased efficiency of public services

Funding programs including AI security National research centres with AI security activities Research for advancing cybersecurity offensive and defensive practices, methodologies and tools Sandboxes, pilots and testing environments for AI security Cyber ranges, simulation platforms and testing environments for secure AI systems Cybersecurity exercises for enhancing AI cybersecurity defences Collaboration for certifying AI systems, training and secure interoperable data exchange

Testing

Promotion of experimentation facilities to test promising AI applications: • Innovation sandboxes • Open piloting and testing environments.

Networking

Collaboration

Enhancement of collaboration opportunities

38