AI135: AI Conformity Assessment in Practice – EU AI Act, VDE SPEC 90012, TÜV AI.AM UPDATE

Agenda:

• Fundamentals and Legal Framework
  • Introduction to AI Systems
    • Types of AI systems and their applications (rule-based, ML, generative AI, hybrid systems)
    • Basic AI architecture and functionality (data collection, training, inference, feedback loops)
    • Distinction: What is an AI system within the meaning of the EU AI Act vs. conventional software?
  • Legal Framework
    • EU AI Act and its implications
    • Risk classification under the EU AI Act
    • What conformity assessments are
    • Conformity assessment procedures for the different levels
    • Definition clarification: minimal risk vs. no risk
    • Requirements for conformity assessment bodies
    • Documentation requirements and technical documentation
    • Declaration of conformity and CE marking
    • Identification and assessment of AI-specific risks
    • Active and passive risk management
  • Data Protection Requirements
    • GDPR relevance for AI systems (Art. 22 – automated decision-making)
    • Data Protection Impact Assessment (DPIA) in the AI context
    • Rights of affected persons in AI-supported decisions
    • Anonymization, pseudonymization, and synthetic data
      
      
• Technical Aspects and Assessment Methods
  • Ethical Assessment of AI Systems
    • Relevance of ethical assessments
    • Conceptual overview
  • Technical Evaluation of AI Systems
    • Assessment of training data and data pipelines
    • Model validation and performance metrics
    • Bias detection and assessment
  • Security and Robustness
    • Cybersecurity for AI systems
    • Adversarial attacks and defense strategies
    • System stability and fault tolerance
  • Quality Assurance and Audit Fundamentals
    • Continuous monitoring
    • Principles and methods of AI auditing
    • Audit trail and traceability
      
      
• Assessment Procedures for AI Systems – Theory & Practice
  • VDE SPEC 90012
    • Structure and objective of the specification
    • Assessment catalog and evaluation criteria
    • Practical application using an example AI system
  • TÜV AI.AM (AI Assessment Matrix)
    • Structure, dimensions, and maturity levels
    • Assessment procedure and scoring
    • Practical application and comparison with VDE SPEC 90012
    • Comparison of both procedures: strengths, weaknesses, application scenarios
      
      
• Examination

Objectives:

Upon successful completion of this course AI135 AI Conformity Assessment in Practice – EU AI Act, VDE SPEC 90012, TÜV AI.AM, participants will be able to:

• Classify AI systems according to the risk categorization of the EU AI Act and identify the relevant conformity assessment procedures
• Correctly apply documentation requirements, declarations of conformity, and CE marking obligations for AI systems
• Take data protection requirements (GDPR, DPIA) into account in the AI context and assess data subject rights
• Technically evaluate AI systems — including training data, model validation, bias detection, as well as security and robustness
• Practically conduct assessment procedures in accordance with VDE SPEC 90012 and TÜV AI.AM and interpret their results
• Compare both assessment methods with regard to their strengths, weaknesses, and appropriate use cases, and apply them situationally
• Methodically plan AI audits, document audit trails, and ensure the traceability of AI-based decisions

Additionally, the course provides a solid foundation for further courses, such as:

• AI100 AI Officer

 

Target audience:

This course AI135 AI Conformity Assessment in Practice – EU AI Act, VDE SPEC 90012, TÜV AI.AM is aimed at professionals and managers who are responsible — or will become responsible — for the development, deployment, assessment, or monitoring of AI systems. The following groups will benefit in particular:

• IT managers and developers who need to design and document AI systems in a compliant manner
• Compliance and data protection officers who want to evaluate AI applications in a legally sound way
• Quality managers and auditors who integrate AI assessment procedures into existing QM processes
• Product owners and AI project managers who are responsible for the entire lifecycle of an AI system
• Consultants and service providers who support organizations in achieving AI compliance

Prerequisites:

No specific certifications are required to participate in this course AI Conformity Assessment in Practice – EU AI Act, VDE SPEC 90012, TÜV AI.AM. However, the following foundational knowledge is recommended to get the most out of the course content:

• Basic understanding of software development or IT systems
• Initial exposure to AI or machine learning (e.g., through practical experience or introductory training)
• Interest in regulatory or compliance-related topics in the IT environment
• Knowledge in the areas of quality assurance, IT security, or data protection is beneficial but not mandatory

Description:

With the EU AI Act, one of the most far-reaching regulations in the field of Artificial Intelligence is now in effect for companies and organizations. Anyone who develops, operates, or evaluates AI systems must be able to demonstrate that these systems meet legal, technical, and ethical requirements — and that is exactly where this course comes in.

The course AI Conformity Assessment in Practice – EU AI Act, VDE SPEC 90012, TÜV AI.AM provides solid knowledge for evaluating and auditing AI systems based on current standards and procedures. From the legal foundations of the EU AI Act and data protection requirements under the GDPR, to specialized assessment methods such as VDE SPEC 90012 and the TÜV AI Assessment Matrix (AI.AM), participants gain a structured and practice-oriented overview of the entire conformity process.

A particular focus is placed on practical application: using concrete example AI systems, both assessment procedures are carried out and directly compared with one another. The course concludes with an examination that certifies the knowledge acquired.