Welche Rolle spielt der Betriebsrat bei KI-Entscheidungen im Unternehmen?

The works council plays an active co-determination role in AI-related decisions within the company, anchored in the Works Constitution Act (BetrVG). As soon as AI systems are deployed that monitor or influence the behaviour or performance of employees, the works council has concrete rights of participation and review. The following sections answer the most important questions about co-determination, information obligations, and works agreements in the context of AI.

What co-determination rights does the works council have regarding AI systems?

The works council has a statutory co-determination right under § 87 Para. 1 No. 6 BetrVG for AI systems used to monitor the performance or behaviour of employees. This right applies regardless of whether monitoring is the primary purpose of the system or arises only as a side effect.

In addition, co-determination rights apply to AI-supported decisions affecting individual employment relationships — for example, automated appraisal systems, AI-supported personnel selection, or algorithmic shift planning. The General Equal Treatment Act (AGG) is also relevant in this context: AI-supported decisions must not be based on characteristics such as gender, age, religion, or ethnic origin. The employer bears full responsibility, even when discrimination arises indirectly through biased training data.

In practice, this means the works council can block the introduction of an AI system or make it subject to conditions until an agreement is reached. This right of veto is a powerful instrument that companies should factor into their AI planning at an early stage.

When must the works council be informed about AI introductions?

The works council must be informed before an AI system that touches areas subject to co-determination is introduced or significantly changed. Informing the works council after the fact is not sufficient in legal terms and can render the introduction measure ineffective.

The information obligation begins early — ideally already in the planning phase, when requirements are being defined and providers evaluated. The works council has a general right to information under § 80 BetrVG that enables it to carry out its tasks. In concrete terms, this means the employer must inform the works council about the type, functionality, purpose, and effects of the planned AI system.

Particularly important is the timely involvement of the works council in high-risk AI systems within the meaning of the EU AI Act, which has been in force since 2 February 2025. These systems are subject to strict requirements, including human oversight mechanisms and information obligations towards those affected. The works council can act as an internal control body in this context.

Which AI applications are subject to co-determination?

In principle, all AI applications that are capable of monitoring, evaluating, or controlling the performance or behaviour of employees are subject to co-determination. The co-determination obligation does not depend on the technical label „AI“, but on the actual function of the system.

Typical examples of AI applications subject to co-determination include:

AI-supported applicant selection and CV analysis
Automated performance appraisal systems
Algorithm-based shift and deployment planning
AI tools for productivity measurement or activity monitoring
Systems for automated dismissal recommendations or promotion decisions
Chatbots or assistance systems that process and evaluate employee data

AI tools that are used exclusively by employees internally, without their use being systematically evaluated, are not automatically subject to co-determination — for example, an AI-supported writing assistance tool without logging. However, the boundary is fluid, which is why a legal review is advisable in individual cases.

How can the works council effectively review AI decisions?

The works council can effectively review AI decisions by demanding transparency about the system’s functionality, training data, and decision-making logic. The basis for this is the right to an explanation of the decision-making process, which the EU AI Act expressly prescribes for automated decisions in high-risk systems.

The following review steps are specifically recommended:

Request technical documentation: How was the system trained, which data sources were used, and how does it make decisions?
Assess discrimination risks: Could training data systematically disadvantage certain groups (AGG review)?
Check human control mechanisms: Is there a human review instance before final decisions are made?
Ensure logging and traceability: Are automated decisions documented and can they be reconstructed?
Inform those affected: Are employees informed about whether and how AI is used in decisions?

For a well-founded review, works councils often need external support from experts, whom they can call upon under § 80 Para. 3 BetrVG. Structured further training on AI fundamentals and legal issues is an important building block for developing the necessary expertise.

What does a works agreement on AI specifically regulate?

A works agreement on AI bindingly regulates the conditions under which AI systems may be used in the company, what rights employees have, and how the works council is involved in ongoing AI processes. It creates legal certainty for both sides.

Typical contents of an AI works agreement include:

Scope: Which AI systems and areas of application are covered?
Transparency obligations: How are employees informed about AI decisions?
Data protection and data access: Which data may be collected, processed, and for how long may it be stored?
Rights of objection: Can employees challenge automated decisions?
Co-determination procedures: How is the works council involved in future AI changes?
Training and qualification obligations: What further training is ensured for affected employees?
Control and audit rights: How and when can the works council review compliance?

A well-drafted works agreement is oriented towards the requirements of the EU AI Act as well as the principles of AI governance as described, for example, in ISO 42001. It not only serves to protect the workforce, but also helps the company to demonstrate AI compliance in a verifiable manner.

What further training do works councils need for AI topics?

Works councils need a combination of basic technical understanding, legal expertise, and practical judgement for AI topics. Without this knowledge, they cannot effectively exercise their co-determination rights when AI systems are introduced.

Specifically, further training for works councils should cover the following areas:

AI fundamentals: What is an AI system, how does it learn, and how does it make decisions? Basic concepts such as machine learning and automated decision-making explained in an accessible way.
Legal framework: BetrVG, AGG, GDPR, and the requirements of the EU AI Act in overview.
Recognising AI risks: Discrimination risks from training data, data protection violations, lack of transparency.
Understanding AI governance: How do AI management systems work, what roles and responsibilities exist within the company?
Shaping works agreements: Which contents are essential, which formulations offer genuine protection?

Works councils are entitled to training that is necessary for their work under § 37 Para. 6 BetrVG. AI topics clearly fall into this category, as they affect the working conditions of the entire workforce. Training should be practice-oriented and include concrete case studies from everyday business life, so that works councils not only acquire theoretical knowledge but become directly capable of taking action.

How mITSM supports works councils and companies on the topic of AI

We at mITSM offer well-founded further training that prepares works councils, managers, and employees for the responsible use of AI. Our trainers are certified experts who not only know AI topics in theory, but convey them from practical experience in companies.

Our offering in the area of AI includes, among other things:

AI Manager training with a focus on AI governance, the EU AI Act, compliance, and risk management
AI Officer according to ISO 42001 for those responsible who want to build or audit an AI management system
Training on fundamentals of AI application for beginners who want to build digital competence and understand AI potential in everyday professional life
In-house training delivered as standardised course formats directly on-site with the customer
Certifications via ICO-Cert as a recognised certification partner for AI topics

Our courses are available as in-person training, online live training, and e-learning, and are aimed at specialists, managers, and companies that want to use AI safely, in a legally compliant manner, and competently. More information about our AI training offering can be found on our website. If you want to check the maturity level of your AI governance, you can also take a look at our ISO 42001 Audit. Get in touch with us and take the next step towards competent AI use.

Dieser Inhalt wurde mithilfe von KI erstellt und kann Fehler enthalten.