The digital transformation of workplace processes and the effective use of AI applications are becoming increasingly critical to companies’ competitiveness and market position. Employers must navigate complex legal requirements, extensive documentation and governance obligations covering both employment law and technology regulation.

Works council and co-determination requirements

Operational change through the introduction of AI

The use of AI does not, in itself, constitute an operational change (Betriebsänderung) under section 111 of the German Works Constitution Act (Betriebsverfassungsgesetz – BetrVG). However, it may do so in particular circumstances, for example where the use of AI or other technology across substantial parts of the business results in a significant reduction in headcount, leading to a fundamental change in the organisation of the undertaking, or introduces fundamentally new working methods.

Whether the introduction of AI amounts to an operational change under section 111 BetrVG will depend on the circumstances of the individual case. In most instances, however, the deployment of AI will merely constitute the further development of existing IT systems without materially affecting operational processes, working practices or employees’ working conditions. Likewise, AI will typically not qualify as a separate “operational facility” (Betriebsanlage), but rather as part of the existing IT infrastructure.

Where the implementation of AI does amount to an operational change, the employer will be required to engage in negotiations with the works council regarding an Interessenausgleich (reconciliation of interests) and, where appropriate, a Sozialplan (social plan).

Participation and co-determination rights of the works council

The deployment of AI systems within an organisation may give rise to a broad range of information, consultation and co-determination rights under German works constitution law. Employers should therefore involve the works council at an early stage and throughout the relevant planning and implementation phases of the project. In addition to its general information and monitoring rights under section 80(1) BetrVG, the works council may appoint an external expert in order to assess the AI systems often complex technical operation and potential impact on employees. Significantly, this right is subject to a lower threshold in the context of AI systems: unlike in other circumstances, the works council is not required to provide a specific justification for the engagement of an expert (section 80(3) sentence 2 BetrVG).

Furthermore, section 90(1) no. 3 BetrVG requires employers to inform the works council, at an early planning stage about working methods and operational processes that are affected by the use of AI. In relation to high-risk AI systems, further disclosure obligations may arise under Article 26(7) AI Act. However, some AI systems fall outside the high-risk systems category by virtue of Article 6(3) AI Act, particularly where their functions are merely preparatory, supportive or otherwise ancillary in nature.

AI literacy

Article 4 AI Act requires employers to ensure that employees possess an appropriate level of AI literacy. This obligation extends to equipping employees with the knowledge and skills necessary to use AI systems responsibly, critically evaluate their outputs, and understand the risks associated with their deployment. The design and implementation of corresponding training measures, including their content, scope and target groups, are subject to the works council’s co-determination rights under section 98 BetrVG. In practice, AI-related training, upskilling and reskilling initiatives are frequently governed by collective arrangements, most notably works agreements or, where restructuring is involved, as part of a reconciliation of interests.

Co-determination rights in practice

Particular practical significance attaches to section 87(1) no. 6 BetrVG. Generative AI systems, AI assistants and intelligent collaboration platforms routinely process usage and log data, including information relating to processing times, communication patterns, file access and the use of particular functions. Such data may enable conclusions to be drawn regarding employees’ conduct, performance or professional capabilities. For the purposes of co-determination, it is immaterial whether the employer intends to monitor employees. According to the settled case law of the German Federal Labour Court, the mere fact that a technical system is objectively capable of monitoring employees’ conduct or performance is sufficient to trigger co-determination rights.

In addition, further co-determination rights may arise, in particular under section 87(1) no. 1 BetrVG (workplace order and employee conduct), nos. 2 and 3 (working time and work organisation), no. 7 (health and safety), and nos. 10 and 11 (remuneration and performance-related assessment systems). Additional participation rights may also be relevant under sections 95 and 99 BetrVG. These rights are supplemented by the protection of employees’ personal rights under section 75(2) BetrVG.

Any failure to comply with applicable co-determination requirements may prevent the introduction or continued use of the relevant AI system. Employers should therefore assess AI applications from a works constitution perspective at an early stage, define their intended purpose clearly and precisely, and maintain comprehensive records of the decisions underpinning their deployment.

Where the use of AI systems affects workplace organisation, performance management or personnel-related decision-making, employers will generally be well advised to conclude a dedicated works agreement. This is particularly relevant in the context of algorithmic management, namely the use of AI systems to allocate tasks, prepare shift rotas, or support decisions relating to remuneration and performance targets. Typical applications include the prioritisation and allocation of work, the preparation of shift and deployment schedules, and the use of AI-supported tools to inform target setting, performance evaluations, bonus payments or other remuneration mechanisms. In such circumstances, co-determination rights may be triggered, in particular under section 87(1) nos. 2 and 3 BetrVG (working time and the allocation of working time) and section 87(1) nos. 10 and 11 BetrVG (remuneration structures and performance-related remuneration systems).

In addition to works constitution requirements, employers must also ensure compliance with applicable data protection legislation and the AI Act. Statutory minimum standards, particularly those arising under the GDPR and the AI Act, cannot be waived or diluted by way of a works agreement. Where disputes arise in relation to matters that are subject to co-determination, employers should also consider at an early stage whether the involvement of a conciliation committee may be appropriate.

By contrast, where AI is deployed as a general productivity or assistance tool, such as through organisation-wide access to AI assistants including Microsoft Copilot, and where the technology does not materially affect employee conduct in a manner giving rise to co-determination rights, a more proportionate approach will often suffice. In such cases, it may be adequate to amend existing IT framework works agreements or introduce an additional AI policy. This enables organisations to establish an appropriate governance framework while maintaining the flexibility necessary for the effective adoption of AI technologies and avoiding unnecessary regulatory complexity.

Data protection and anti-discrimination law

AI systems regularly process personal data relating to employees and job applicants, for example in the context of automated CV screening, evaluating qualifications, assessing cultural fit, or predicting performance and potential. Employers must therefore ensure that such processing is supported by a valid legal basis and complies with the fundamental data protection principles of purpose limitation, data minimisation and transparency.

A works agreement may serve as a legal basis pursuant to section 26(4) Federal Data Protection Act (Bundesdatenschutzgesetz – BDSG). Alternatively, employers may rely on employee consent under section 26(2) BDSG. In the employment context, however, in light of the inherent imbalance of power between employer and employee the voluntary nature of consent is often open to doubt (see German Data Protection Conference (DSK), Short Paper No. 14 of 24 September 2020). Under section 26(2) BDSG, consent may nevertheless be considered voluntary where employees receive a legal or economic benefit, or where the parties pursue genuinely aligned interests. In practice, consent is therefore more likely to be relevant in relation to optional employee benefits than to processing activities that are intrinsically linked to the employment relationship.

Depending on the nature, scope and risk profile of the intended processing, employers may also be required to conduct a data protection impact assessment pursuant to Article 35 GDPR. This will be particularly relevant where AI systems involve extensive profiling, automated decision-making, or other forms of processing that are likely to result in a high risk to the rights and freedoms of employees or applicants.

Where AI systems are deployed in HR processes, such as the pre-selection of applicants or the evaluation of employee performance, they frequently generate automated assessments of personal attributes, including suitability, performance and development potential. From a data protection perspective, these activities will generally constitute “profiling” within the meaning of Article 4(4) GDPR.

This gives rise to a range of legal and compliance risks, particularly in relation to algorithmic bias and indirect discrimination. For example, AI systems may disadvantage certain groups of applicants or employees on the basis of protected characteristics such as age, sex or ethnic origin. Under the burden-of-proof regime under section 22 General Equal Treatment Act (Allgemeines Gleichbehandlungsgesetz – AGG), the burden shifts to the employer once an employee or applicant establishes facts capable of giving rise to a presumption of discrimination. Employers must therefore be able to demonstrate that the AI systems they deploy operate in a fair and non-discriminatory manner and that employment-related decisions are based on objective, transparent and verifiable criteria.

To the extent that AI systems materially influence personnel decisions, employers must also have regard to the restrictions on automated decision-making set out in Article 22 GDPR. Particular care is required where AI-generated outputs are used as a basis for decisions that produce legal effects or similarly significant consequences for employees or applicants.

Failure to comply with data protection regulations may not only expose employers to substantial regulatory sanctions but may also cast doubt on the admissibility of the relevant processing activities under employment law. Employers should therefore undertake a thorough assessment of data flows, risks and appropriate mitigation measures before deploying AI systems and ensure that compliance with applicable data protection requirements can be demonstrated at all times in accordance with the accountability principle enshrined in Article 5(2) GDPR.

Use of AI in employment decisions

The European legislator has expressly recognised the particular risks of using AI in employment-related decision-making. Pursuant to Annex III, section 4 AI Act, AI systems used for the recruitment and selection of applicants, promotion decisions, dismissals, task allocation, or the monitoring and evaluation of employee performance are, in principle, classified as high-risk AI systems. Such systems are subject to extensive regulatory requirements, including obligations relating to risk management, data governance and quality, transparency, human oversight, technical documentation and post-market monitoring. Organisations intending to deploy such systems should therefore familiarise themselves with the applicable compliance framework at an early stage.

As set out above, as deployers of high-risk AI systems, employers are required to inform employee representatives and affected employees prior to the system’s deployment (Article 26(7) AI Act). In addition, the AI Act mandates effective human oversight (Article 14 in conjunction with Article 26(2) AI Act). Employment-related decisions must not, in practice, be delegated to an AI system. Decisions concerning recruitment, promotion, transfers, disciplinary measures or dismissals must ultimately remain subject to meaningful human review and decision-making. Employers should therefore establish and document appropriate governance, oversight and decision-making processes in accordance with Article 26(6) AI Act.

Certain deployers, in particular public authorities and private entities providing public services, may also be required to conduct a Fundamental Rights Impact Assessment pursuant to Article 27 AI Act. This assessment involves a structured evaluation of whether, and to what extent, the intended use of an AI system could adversely affect the fundamental rights of affected individuals, including the rights to equal treatment, data protection and procedural fairness. Its purpose is to identify and address such risks before the system is deployed and to ensure that appropriate safeguards are implemented from the outset.

Non-compliance with the AI Act may result in regulatory enforcement measures and substantial administrative fines of up to EUR 15 million or 3% of the undertaking’s total worldwide annual turnover, whichever is higher. However, the specific obligations applicable to high-risk AI systems used in the employment context will not become fully applicable until 2 August 2027. Employers should use the intervening period to review existing HR processes, governance frameworks and compliance structures, and to take the necessary preparatory steps to ensure timely compliance with the new regulatory requirements.

Individual employment law limits

Scope of the employer’s managerial prerogative

Pursuant to the employer’s managerial prerogative under section 106 German Trade Regulation Act (Gewerbeordnung – GewO), employers may generally determine whether, to what extent and for what purposes AI systems are used in the workplace. Equally, employers may prohibit the use of specific AI applications, particularly tools that have not been approved for business purposes or that pose legal, security or compliance risks.

The introduction of such requirements should be accompanied by clear governance frameworks, usage policies, training measures and compliance standards. Employers should also ensure that employees receive appropriate guidance on data protection obligations, the handling of confidential information and the risks associated with AI-assisted decision-making, taking into account the particular use case and risk profile of the relevant system. This is of particular importance because breaches of the AI Act or applicable data protection legislation may be attributable to the employer as the deployer or operator of the AI system.

The employer’s authority is limited by the employee’s contractual obligation to perform work personally. Employees may not delegate all of their contractual duties to AI systems or use AI-generated outputs without exercising appropriate professional judgement and review. This applies particularly to tasks requiring specialist expertise, independent judgement, creative input or autonomous decision-making. Failure to comply with these obligations may give rise to employment-law consequences, including disciplinary measures and, in serious cases, dismissal.

AI systems are also increasingly being used to support the management and organisation of workplace processes. However, the employer’s managerial prerogative cannot itself be delegated to an AI system. While AI may assist in preparing, informing or recommending decisions, responsibility for instructions, management decisions and personnel measures must remain with a human decision-maker at all times. This applies in particular to performance evaluations, workforce scheduling, promotions, disciplinary measures and decisions relating to the hiring or termination of employees.

Conclusion

The use of AI in the workplace is no longer a matter of future speculation; it is now central to workplace compliance, digital transformation and workforce management initiatives. Employers must therefore address the relevant employment law, works constitution law, data protection and AI regulatory requirements as AI is deployed. Particular emphasis should be placed on the early involvement of employee representatives, establishing robust governance and oversight frameworks, and carefully documenting AI-enabled processes and employment-related decisions.

Practical take away: Employers should make effective use of the transitional periods provided under the AI Act to identify and review existing and planned AI applications, classify them according to the applicable risk categories, and implement the necessary compliance measures. Organisations that address AI governance proactively, establish clear lines of accountability and invest in workforce readiness at an early stage will be best placed to navigate the evolving regulatory landscape while unlocking the operational benefits of AI in a legally compliant and sustainable manner.

Contact