Information and consultation on artificial intelligence in the workplace

Introduction

The use of artificial intelligence (AI) in the workplace is growing rapidly. More and more employers are recognizing the benefits and investing in it. European regulators saw this (r)evolution coming and are attempting, through the AI Regulation (Regulation (EU) 2024/1689 of the European Parliament and of the Council of June 13, 2024), to strike the right balance between AI innovation and fundamental rights in a timely manner.

As a reminder: the AI Regulation has been in force since August 1, 2024, and the obligations are being phased in gradually. For example, the prohibitions on AI systems posing an unacceptable risk (e.g., social scoring systems that assess based on social behavior or personal characteristics and may lead to discrimination) are already in effect. Most rules relevant to AI systems in the workplace (classified as high-risk AI systems) are set to take effect in August 2026 (see below, however). This includes, for example, AI systems used to post targeted job openings, analyze and filter applications, and evaluate candidates.

Despite the increase in use and the imminent (further) entry into force, last year’s Littler European Employer Survey revealed that the majority of employers do not feel prepared for the obligations imposed by the AI Regulation.[1]

Many employers likely breathed a sigh of relief when, on March 26, 2026, the European Parliament approved its negotiating position on the AI section of the Digital Omnibus.[2] With this “Digital Omnibus,” Europe is seeking to simplify existing EU digital regulations. For instance, regarding AI, it is proposed to postpone the entry into force of the rules for high-risk AI systems (concerning technical and organizational measures and information (and consultation)). Simplification is also on the table regarding AI literacy, particularly the responsibility of employers to adequately train employees who work with AI systems.

Can employers then ease up on their preparations? That is not advisable, because until these proposals are accepted, the current rules remain in effect. Moreover, most obligations will come into effect regardless (with or without a delay). As we wrote in an earlier LinkedIn post, it is therefore important to start mapping out the AI systems in use right now.

Furthermore, it should not be forgotten that, even aside from the AI Regulation, there are already numerous obligations in Belgium regarding the information and consultation of employees (and their representatives). Employers would be wise to determine now which bodies within the company, where necessary, will be informed (and consulted). In this Insight post, we discuss the main information and consultation obligations applicable to AI in the workplace. 

The Works Council

AI and Collective Bargaining Agreement No. 9

In general, pursuant to Article 9 of Collective Bargaining Agreement No. 9 of the National Labor Council, the employer is required to inform the works council in advance of any measure that may alter aspects of personnel policy. This includes, for example, measures concerning recruitment and selection, as well as career advancement. An AI system used to screen job applications or evaluate employees would therefore quickly fall under this category.

Employers are also required to inform the works council of measures that may alter working conditions and terms of employment within the company or a department.[3]

In addition, the works council must also be informed in advance and consulted if decisions could entail significant changes to the organization of work or employment contracts.[4] In Belgium, a cascade system applies in this regard: if there is no works council, the union delegation takes over. If there is neither a works council nor a union delegation, the Committee for Prevention and Protection at Work (CPBW) acts in place of these bodies.

In general, it can be concluded that AI systems implemented on a large scale quickly fall under these definitions. 

AI as a new technology within the meaning of Collective Bargaining Agreement No. 39

Furthermore, if an AI system qualifies as a “new technology” within the meaning of Collective Bargaining Agreement No. 39 of the National Labor Council, an employer who typically employs more than 50 workers (to be assessed in the calendar year preceding the introduction of the new technology) must take specific prior action.[5]

In that case, the obligation applies to share, 3 months prior to the introduction of the “new technology”:

• Share written information regarding the nature of the new technology, the factors justifying its introduction, and its social consequences;
• Consult with the works council, or, if there is no works council, with the union representatives, regarding the social consequences of the new technology.

However, experience shows that this collective bargaining agreement has fallen into disuse.[6] One reason for this is likely the fact that “new technology” is only considered when it has significant collective consequences for employment, work organization, or terms and conditions of employment. Such collective consequences only apply when 50% and at least 10 employees of a specific occupational category are involved.[7] As a result, employers who implement an AI system only incrementally (each time affecting less than 50% of an occupational category) often fall outside the scope of application.[8]

Nevertheless, it is advisable to conduct the assessment. After all, an employer who does fall within the scope of application and fails to comply with the obligations cannot subsequently unilaterally terminate an employee’s contract, except for reasons unrelated to the introduction of the new technology. If the employer does so anyway, they risk a lump-sum compensation equal to three months’ gross salary. If an AI system is introduced with the specific purpose of evaluating employees and potentially deciding on the termination of the employment relationship, this would, of course, undermine the very purpose of the AI system.

The duty to inform is, however, limited and does not concern the individual labor law consequences of introducing new technology, but rather more general matters such as:

• Who is qualified to work with the new technology and who requires training;
• The health and safety of employees;
• Etc.

Although Collective Labor Agreement No. 39 is no longer in use and certain companies will not fall within its scope, the information obligations need not necessarily be viewed as an unnecessary burden. Most of these obligations will have to be fulfilled by companies using AI anyway. For example, the obligation to inform and consult on “who is qualified to work with the new technology” is virtually identical to the AI literacy requirement in the AI Regulation. Likewise, the obligation to inform about the impact of the new technology on “the health and safety of employees” is, in any case, part of an employer’s general duty of care.

The union delegation

When an AI system could lead to a change in contractual and customary terms and conditions of employment, Collective Bargaining Agreement No. 5 of the National Labor Council (often transposed into a sectoral equivalent) requires the employer to inform the union delegation prior to implementation.[9] For example, an employer wishing to introduce an AI system for assigning tasks based on individual behavior or personal traits or characteristics will thus be required to inform the union representatives.[10]

Apart from the specific information obligations, the fact is that when an AI system requires a significant change to agreed-upon work and job duties on a larger scale, reaching a separate agreement with each affected employee may not be a realistic option. Situations in which the employer will be required to negotiate new job classifications and conclude a collective bargaining agreement on this matter are therefore not inconceivable.[11]

The Committee for Prevention and Protection at Work

AI systems can also impact employee well-being and are therefore not separate from employers’ well-being obligations.[12] Consider, for example, a warehouse where wearables are used to measure employee productivity based on certain parameters and rank their performance. An employer must implement a well-being policy, develop a dynamic risk management system, and take the necessary preventive measures.

When developing, planning, implementing, and evaluating the dynamic risk management system, the comprehensive prevention plan, and the annual action plan, the employer must consult the CPBW. The CPBW has the general mandate to issue prior advice and formulate proposals in this regard.[13] This advisory authority is broad and applies to “all proposals, measures, and resources to be implemented that may, directly or indirectly, immediately or over time, affect the well-being of employees in the performance of their work.”[14]  If no CPBW has been established, these tasks are carried out by the union delegation or, if there is no such delegation, by the employees themselves.[15] The impact of AI on employee well-being therefore, in principle, requires prior consultation with the CPBW.[16]

Conclusion

When an employer is seriously considering implementing an AI system, the question arises as to which obligations regarding social consultation apply. The nature of the AI system and its impact on terms and conditions of employment, working conditions, and employment contracts, as well as the social consultation bodies present within the company, determine the information and consultation procedure to be followed.

However, these obligations need not be viewed as a burdensome sum of requirements. After all, there is significant overlap between the obligations regarding information and consultation, the well-being obligations, and the upcoming requirements of the AI Regulation, making an integrated approach advisable. 

In conclusion, it can be said that, although certain obligations under the AI Regulation may be postponed, employers would be well advised to identify the AI systems they currently use or plan to use, gain insight into future applications, and determine which information and consultation procedures must be followed. After all, the use of AI and the associated obligations will not disappear and are inextricably linked.

[1] Littler’s European Employer Survey 2025, 2025_littler_european_employer_survey_report.pdf.

[2] Proposal for a REGULATION OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL amending Regulations (EU) 2016/679, (EU) 2018/1724, (EU) 2018/1725, (EU) 2023/2854 and Directives 2002/58/EC, (EU) 2022/2555, and (EU) 2022/2557 as regards the simplification of the digital legislative framework, and repealing Regulations (EU) 2018/1807, (EU) 2019/1150, (EU) 2022/868, and Directive (EU) 2019/1024 (Digital Omnibus).

[3] Art. 10 of Collective Agreement No. 9.

[4] Art. 4, paragraph 3, of Collective Agreement No. 9.

[5] Collective Bargaining Agreement No. 39, December 13, 1983, concerning information and consultation on the social consequences of the introduction of new technologies, Belgian Official Gazette February 8, 1984.

[6] NISSEN, L., “Working in a digitized world: the need for collective consultation on changes to terms and conditions of employment and the obligation to undergo further training?” in DELANOTE, M., PEETERS, B. and VAN DE WOESTEYNE, I. (eds.), Digitalization. XLVIth Postgraduate Cycle Willy Delva, Mechelen, Wolters Kluwer Belgium, 2021, 181–206.

[7] Art. 2 of Collective Bargaining Agreement No. 39; Employees considered to be affected are those who either must use the new technology under modified terms and conditions of employment or must endure its consequences in terms of employment opportunities as a result of dismissal or transfer; The 50% and the 10 employees are not calculated separately for each occupational category, but for all occupational categories combined in which a change in employment, work organization, or terms and conditions of employment occurs as a result of the introduction of the new technology, when these occupational categories together comprise fewer than 100 employees.

[8] However, this should be nuanced for smaller companies: the 50% threshold and the 10 employees are not calculated separately for each occupational category, but for all occupational categories combined in which a change in employment, work organization, or working conditions occurs as a result of the introduction of the new technology, when these occupational categories together comprise fewer than 100 employees.

[9] Art. 14 of CLA No. 5.

[10] Art. 14 of CLA No. 5.

[11] NISSEN, L., “Working in a digitized world: the need for collective consultation on changes to terms and conditions of employment and the obligation to provide further training?” in DELANOTE, M., PEETERS, B. and VAN DE WOESTEYNE, I. (eds.), Digitalization. XLVI Postgraduate Cycle Willy Delva, Mechelen, Wolters Kluwer Belgium, 2021, 181–206.

[12] Art. 5 of the Workplace Well-being Act.

[13] Art. II.7-2 of the Workplace Well-being Code

[14] Art. II.7-3,1° of the Well-being at Work Code.

[15]Art. II.7-1 of the Workplace Well-being Code.

[16] Art. II.7-2 and II.7-3 of the Workplace Well-being Code.