AI and Works Councils: Shaping the Rollout Together

When organizations discuss AI adoption, the conversation is often about models, platforms, and use cases. When works councils discuss AI, it is about working conditions, data protection, and co-determination. Both perspectives are legitimate. And both are too rarely brought together.

The result: projects that are technically ready get stuck in alignment. Or worse: they launch without the works council, and conflicts come later: more expensive and with more friction.

What the law provides

German labor law gives works councils clear rights regarding AI adoption. The relevant provisions are found in the Works Constitution Act and primarily cover three areas.

Right to information. The employer must inform the works council in advance about planned changes involving new technologies. This includes AI. In advance means: before the decision is made, not after.

Co-determination regarding monitoring. When an AI system is capable of monitoring employee behavior or performance, co-determination applies. This affects obvious cases like performance dashboards as well as any other tool that collects usage data. And most AI systems do.

Engaging experts. Since the amendment to the Works Constitution Act, works councils may engage external experts when evaluating AI systems. The legislator recognizes that AI is new territory for works councils, requiring technical understanding.

Beyond these core rights, additional provisions apply depending on the use case. When AI is used in recruiting, co-determination rights regarding selection guidelines come into play. When AI fundamentally changes work processes, it may constitute an operational change. And when AI systems shift qualification requirements, the works council has a say in vocational training.

Why the works council is not the obstacle

In practice, many organizations experience the works council as a brake. The AI solution is ready, teams want to start, and then an alignment process begins that takes weeks or months. This is frustrating: but usually self-inflicted.

The works council becomes an obstacle when it is involved too late. When it faces accomplished facts. When it receives information only on request. When it feels that its role is a formality.

Works councils that are involved early, that understand the benefit of a solution, and that know their concerns are heard, almost always work constructively. Because they see the same advantage.

It is worth understanding the dynamic: works councils that feel blindsided respond with caution and resistance. This is not obstruction. It is a predictable protective reaction. Organizations that recognize this make negotiation more productive.

What a framework agreement can achieve

Many organizations negotiate a separate works agreement for every single AI tool. This is costly, slow, and creates a patchwork of regulations that becomes confusing over time.

An AI framework agreement solves this problem. It defines fundamental rules for the use of AI in the organization without having to negotiate every single tool separately. Such an agreement typically contains provisions on data protection and data processing, transparency toward employees, classification of AI systems by risk level, and processes for introducing new tools.

The key benefit: low-risk systems, such as AI-powered translation software: can be introduced under the framework agreement without requiring a new negotiation each time. Higher-risk systems, such as AI in recruiting: go through a defined review process.

A good framework agreement is not a rigid document. It should include mechanisms that allow adaptation to new developments, such as regular review cycles or a clearly defined escalation process for new categories of AI systems. Technology evolves faster than any agreement. Planning for this avoids renegotiations.

The works council perspective

It helps to switch perspectives. Works councils face a dual challenge. On one side, they represent employee interests. And those are diffuse when it comes to AI: some worry about their jobs, others about surveillance, others about fairness in automated decisions. On the other side, many works council members lack the technical knowledge to assess AI systems with confidence.

Organizations that understand this offer their works council training. As a strategic investment. A works council that understands what a language model does and what it does not can make informed decisions. And that accelerates the process for everyone involved.

At the same time, organizations should invite the works council to work with the tool themselves. Those who know AI only from slides and data sheets judge it differently than someone who has experienced it in practice. The best way to address concerns is not a presentation. It is experience.

The EU AI Act as common ground

The European AI regulation creates a framework that offers orientation for both sides. The EU AI Act classifies AI systems by risk level and sets clear requirements for transparency, documentation, and human oversight. This affects employers directly: but it also gives works councils a vocabulary and structure for evaluating AI rollouts.

Organizations that take the EU AI Act seriously automatically create a basis for works council engagement. Because what is required by regulation largely overlaps with what works councils demand anyway: transparency, documentation, and clear accountability.

A practical path: building trust

In practice, an approach that proceeds in several steps tends to work well. First, the works council is invited into the strategic discussion: as a co-creator. Then, both sides jointly define which types of AI systems carry which risks. Building on that, a framework agreement is developed that leaves room for change.

In parallel, the works council is given the opportunity to test the platform themselves. This builds trust at a level that no negotiation can replace: personal experience. Organizations that take this path report shorter negotiation times and more stable agreements. Because it can speak with authority.

Involve early, roll out faster

Experience shows a clear pattern: organizations that include the works council from the start of their AI strategy deploy faster than those that seek the conversation only after the fact. Because negotiations are more constructive.

This works especially well when the AI solution itself supports transparency: through audit logs, usage dashboards, and role-based access controls. Because then the works council does not have to trust. It can verify.

Documentation as a shared language

A practical problem that arises in many negotiations between works councils and employers is the lack of a shared language. IT talks about models, APIs, and tokens. The works council talks about co-determination, employee protection, and transparency. Both often mean the same thing: but they do not understand each other.

Technical documentation that makes clear what an AI system does, what data it processes, and how decisions are made bridges this gap. As a factual description that both sides can read and understand.

Platforms that generate such documentation automatically: overviews of models used, data types processed, and access permissions, for example: significantly ease the dialogue. They spare both sides the effort of laboriously gathering information and create a factual basis for constructive negotiation.

Co-determination as a quality marker

Organizations that view co-determination as a quality marker win in the long term. Because an AI system that survives the works council's scrutiny is usually a better system. The questions a works council asks: What happens to the data? Who is liable for errors? How transparent is the system?: are the same questions every organization should ask itself.

Organizations that answer these questions proactively do not just build trust with the works council. They build a robust governance framework that also holds up before customers, regulators, and their own workforce.

From individual solutions to AI strategy

Involving the works council becomes most valuable when it happens at the level of overall strategy. A works council that is only consulted about individual applications can only assess point by point. A works council that is involved in the strategic discussion can help shape the whole picture.

This means: jointly defining how AI should fundamentally be used in the organization. What principles apply? Where are the red lines? What processes ensure that new developments do not bypass co-determination?

Organizations that take this step create a foundation that extends far beyond the current tool landscape. They build a partnership that can adapt to new technologies: because the foundation is right.