ChatGPT as a work tool

Artificial intelligence (AI) is increasingly becoming an integral part of modern working life. The possibilities for its application seem almost limitless.

16 July 2025

Publication

Loading...

Listen to our publication

0:00 / 0:00
  • „AI will not replace you. People who use AI will“ -

Artificial intelligence (AI) is increasingly becoming an integral part of modern working life. The possibilities for its application seem almost limitless. Examples range from the use of AI-controlled robots in manufacturing processes to the use of chatbots in customer service. The text generator ChatGPT (chat generative pretrained transformer) is particularly widely used and now reportedly boasts more than 500 million weekly users. ChatGPT was trained with large amounts of data to understand and generate language. It is therefore able to give human-like responses to text input and even write texts itself. Its data comes from various internet sources and its own training data, which was generated by user input, among other things. ChatGPT is versatile and can be used, for example, as a research tool, translator, for data analysis, programming or even in business correspondence (e.g. when writing emails or letters of termination).

Due to its wide range of uses as a tool in everyday working life, various questions arise from an employment law perspective regarding the use of the chatbot, in particular to what extent employers can instruct their employees to use ChatGPT when performing work or prohibit its use (I.), whether employees may perform their work with the help of the chatbot in the absence of instructions from their employer (II.), and in which constellations the participation rights of the works council must be observed (III.).

We have already reported several times on the new AI Regulation of the European Union (AI Regulation) applicable in this regard (AI in Employment Law | Simmons & Simmons). It should be noted that, in addition to the AI Regulation, the familiar German labour law provisions continue to apply (see, for example, Art. 2 para. 11 AI Regulation for co-determination rights or Recital 63 of the AI Regulation for the General Data Protection Regulation (DS-GVO)).

I. Can employers instruct their employees (not to) use ChatGPT?

The employer's right to issue instructions (Sec. 106 of the German Trade Regulation Act (GewO)) includes the use of certain work equipment, which is why the use of ChatGPT to perform tasks can, in principle, be both ordered and prohibited within the scope of reasonable discretion. This is particularly true because the information entered into ChatGPT is regularly transmitted to the operator and used to train the AI, which in principle can also lead to other users gaining access to the data. There is therefore a risk of disclosure of confidential data, trade secrets and the unlawful processing of personal data (Art. 4 No. 1 DS-GVO), so that a legitimate interest on the part of the employer is unlikely to be denied. In addition, the use of AI systems to generate work results entails intellectual property risks (e.g., a work created using AI may not enjoy copyright protection or third-party copyrights may be infringed).

To clarify the extent to which employees are permitted to use ChatGPT, employers may consider introducing appropriate usage guidelines. These may include, for example, specifications on the tasks for which AI may be used, the extent to which work results may be generated by AI, the cases in which employees may rely on the work results of an AI system, and the information and data that may be fed into AI systems. In addition, employees should be informed about the risks associated with the use of AI systems. Furthermore, employee training on how to deal with AI is usually essential. The new AI Regulation now makes such training mandatory (see Art. 4).

If employees violate their employer's instructions regarding the use of AI, consequences under employment law such as warnings and dismissals are possible. In addition, the unauthorised or improper use of AI systems may give rise to claims for damages by the employer (e.g. if research results from ChatGPT are adopted completely unchecked). In this respect, the principles established by the Federal Labour Court (BAG) on limited employee liability, which make the extent of liability dependent on the degree of fault, apply.

II. Are employees allowed to perform their work using ChatGPT?

If the employer has not given permission, the question arises as to whether employees may use ChatGPT on their own initiative.

In case of doubt, the work must be performed by the employee in person (Sec. 613 para. 1 German Civil Code (BGB)). It therefore constitutes a highly personal obligation which, in principle, may not be delegated to other persons. However, if an employee performs a significant part of their work with the help of ChatGPT (e.g. a programmer has ChatGPT write a code), it should be sufficient for the assumption of personal performance if the employee carefully checks the work results provided by ChatGPT and, if necessary, makes improvements in some areas. In the vast majority of cases, ChatGPT would then be considered a mere tool.

Employees may therefore use the chatbot in principle, provided that the employer has not prohibited this, that this does not violate any other obligations (e.g. protection obligations under Sec. 241 para. 2 BGB) towards the employer and that at least a certain amount of their own contribution remains in the work results created. As the question has not yet been clarified by case law, employers should avoid uncertainties from the outset and specify the extent to which work may be performed with the aid of ChatGPT or AI systems in general.

If employees perform their work with the help of ChatGPT, they are required under their ancillary obligations under their employment contract (Sec. 241 para. 2 BGB) to disclose this use even without being asked to do so by their employer. Such an obligation is likely to exist in any case if the specific type of use could have serious negative consequences for the employer (e.g. due to the disclosure of highly sensitive information). In addition, employers are generally entitled to a right to information (based on Sec. 242 BGB) from their employees as to whether and to what extent they use ChatGPT for their professional activities. Accordingly, such information may, in principle, be requested.

III. When does the works council have participation rights?

If there is a works council, the question arises as to whether it has participation rights with regard to the use or introduction of ChatGPT.

Introduction of technical equipment (Sec. 87 para. 1 No. 6 BetrVG)
The introduction and use of technical equipment (including AI systems) intended to monitor the behaviour or performance of employees is subject to co-determination in accordance with Sec. 87 para 1 No. 6 Works Constitution Act (BetrVG). According to the established case law of the Federal Labour Court (BAG), it is sufficient if the monitoring is objectively and directly suitable, regardless of whether the employer pursues this goal and also evaluates the data obtained through monitoring. This broad interpretation of the term is likely to mean that many AI applications will only be possible with the consent of the works council.

However, with regard to the use of ChatGPT, the Hamburg Labour Court ruled last year (decision of 14 January 2024 – 24 BVGa 1/24) that permission to use AI tools, accompanied by the publication of guidelines and a manual on how to use them, is not subject to co-determination if the AI tools are not installed on the employer's computer systems and employees can only use the AI tools via their private accounts using a web browser. Sec. 87 para. 1 No. 6 BetrVG requires that the information obtained about the behaviour and performance of employees by means of the technical equipment introduced must remain available for at least a certain period of time and be accessible to the employer. In the case in question, the latter was not possible because the employer had no way of knowing whether and to what extent employees were using ChatGPT, so that the assumption of monitoring pressure characteristic of the right to co-determination did not exist.

Order in the workplace and conduct of employees (Sec. 87 para 1 No. 1 BetrVG)
With convincing arguments, the labour court also rejected a right of co-determination under Section 87 (1) No. 1 BetrVG, which applies to questions of the organisation of the business and the conduct of employees. Requirements for the use of ChatGPT (in this case in the form of guidelines and a manual) do not concern the organisation of the business or the conduct of employees, but rather the manner in which work is performed, which is why they fall under the employer's right to issue instructions (Sec. 106 GewO) and not under a co-determination provision.

Planning of work processes and workflows (Sec. 90 para. 1 No. 3 BetrVG)
However, the labour court left open the question of whether the participation rights under Sec. 90 para. 1 No. 3 BetrVG were relevant. According to this provision, the employer must inform the works council of the planned collective use of artificial intelligence in work processes and workflows, submitting the necessary documents. Furthermore, the employer must consult with the works council on the planned measures and their effects on the employees (Sec. 90 para. 2 BetrVG).

Personnel selection guidelines (Sec. 95 para. 2a BetrVG) and personnel questionnaires and assessment principles (Sec. 94 BetrVG)
The establishment of collective personnel selection guidelines using artificial intelligence requires the consent of the works council in accordance with Sec. 95 para 2a BetrVG. The same applies to the use of personnel questionnaires and assessment principles (Sec. 94 BetrVG). The term "personnel questionnaire" covers all formalised and standardised information gathering by the employer on employee data (in particular employee surveys). Assessment principles are rules that serve to assess the performance and conduct of employees and apply to a specific group of employees. The right to co-determination would therefore have to be taken into account, for example, if ChatGPT designs personnel questionnaires or compares the performance of employees according to criteria set by the employer.

Operational changes (Sec. 111 et seq. BetrVG)
Finally, it is also conceivable that a change in operations subject to co-determination may exist if the use of AI systems in a company has such a significant impact on work processes or the organisation of a business that it leads, for example, to the closure of part of the business, a change in operating facilities, or fundamental changes to the business organisation, the purpose of the business or the operating facilities. In order to answer the question of whether a fundamental change has taken place, it is crucial to determine whether significant disadvantages for the employees affected are to be expected.

Consultation of an AI expert (Sec. 80 para. 3 sentence 2 BetrVG)
Finally, it should also be noted that the works council may, in agreement with the employer, consult experts in the performance of its duties to the extent necessary for the proper performance of its duties (Sec. 80 para. 3 sentence 1 BetrVG). If the introduction or application of artificial intelligence has to be assessed, the involvement of an expert is often considered necessary (Sec. 80 para. 3 sentence 2 BetrVG). In this case, however, the prerequisite for the application of the provision is that the tasks of the works council are actually affected and that the works council has to assess the introduction and application of AI. The legislator estimates the average cost of an expert at EUR 833 (BT-Drucks. 19/28899, p. 17).

Implementation of company training measures (Sec. 98 BetrVG)
Employee training courses are subject to the co-determination rights under Sec. 98 BetrVG as internal training measures. This also applies to AI training courses, which are now mandatory under the AI Regulation but are also often recommended independently of this.

IV. Summary

The employer can, in principle, both order and prohibit the use of ChatGPT for the performance of tasks by exercising its right to issue instructions. If they do not provide their employees with guidelines on how to use ChatGPT, they may only use the chatbot to the extent that this does not violate any other obligations (e.g. protection obligations under Sec. 241 para. 2 BGB) towards their employer and at least a relevant share of the work result created remains. Employers may require their employees to disclose whether they use ChatGPT to perform their tasks. When implementing ChatGPT in everyday work, the works council's co-determination rights may have to be observed on a case-by-case basis.

This document (and any information accessed through links in this document) is provided for information purposes only and does not constitute legal advice. Professional legal advice should be obtained before taking or refraining from any action as a result of the contents of this document.