Collective bargaining on artificial intelligence at work

At European level, the Friedrich Ebert Stiftung published a report(opens in new tab)This link opens in a new tab on collective bargaining on AI in European services sectors in December 2024. It was based on an analysis of 31 collective bargaining agreements that mention AI and a survey of trade union representatives from trade unions affiliated to Uni Europa in 32 countries. According to the report, only 20%(opens in new tab)This link opens in a new tab of the trade unions reported having a collective bargaining agreement that addressed AI at the organisation or sector level, while 42% indicated that they were engaged in some social dialogue discussions or negotiations related to AI issues.

Also at European level, but in the metal, engineering and technology (MET) sector, a two-year iMET project is running from April 2024 to March 2026, enabling innovative social dialogue and collective bargaining on AI. The project has the joint support of project partners at the European level, including CEEMET and industriAll Europe(opens in new tab)This link opens in a new tab (associated partners), as well as employer organisations in Croatia, Latvia, Poland, Serbia and Spain, and trade union federations in Austria, Croatia, France, Poland, Portugal, Romania, Serbia and Spain with additional expert support from the University NOVA Lisbon. The iMET project is focused on improving social partners’ capacities to be informed, consulted and engaged in the digital transformation of MET sector companies and to monitor these effects on employment, job requirements and working conditions. For social partners, it is particularly relevant that iMET activities address the impact of digitalisation and AI on jobs, especially regarding the impact of AI on occupational health and safety; human resource management; data protection; and skills needs. The project will produce e-manuals, tutorials, creative videos and publications. It will involve over 900 participants in European and national workshops.

In the banking sector, a joint declaration on AI(opens in new tab)This link opens in a new tab was agreed between the European Banking Federation (EBF) and UNI Europa on 13 March 2024. A similar declaration had already been agreed in the insurance sector(opens in new tab)This link opens in a new tab on 14 March 2021. These European initiatives have inspired and facilitated similar developments at national level.

A number of organisations represent the interests of the AI industry at European level: Digital SME Alliance(opens in new tab)This link opens in a new tab has members in 6 Member States and Digital Europe(opens in new tab)This link opens in a new tab has affiliated national associations in 24 Member States. At national level, for example, AMETIC has established Spain’s first Artificial Intelligence Observatory. Its aim is to increase knowledge and use of AI in Spain. This observatory analyses AI’s impact on areas such as legislation, ethics, cybersecurity, the economy and privacy.

Examples of collective bargaining on AI were reported in Italy at cross-industry level, in Belgium and Ireland at sector level, and in Denmark at company level. In Spain, there are examples of collective bargaining on AI at all three levels.

The Italian cross-industry agreement of 15 April 2024 amended the national trade collective agreement (CCNL(opens in new tab)This link opens in a new tab) in force, addressing AI-related issues. The agreement envisages a broadening of the scope of application of the CCNL to include, for example, design, graphics and planning companies; garages and car repair workshops; and an important innovation, companies dealing with AI. The CCNL also recognises the need to introduce positions such as senior AI management and coordination officers and AI ethics and responsibility experts to manage and coordinate the introduction of AI tools in companies in a constructive and responsible manner. They should also be able to assess the impact that AI might have in a company, minimising the risk of worker substitution for the tasks most affected and trying to utilise human resources as effectively as possible. The CCNL emphasises the importance of information, education and training in the implementation and/or use of AI systems in order to reskill the resources already present in the workforce. One way it can do this is through the use of tools provided by the bilateral body EBIASP, in particular by extending the use of the ‘active notice voucher’, already aimed at retraining human resources.

In Belgium, collective bargaining talks took place in the banking and insurance sector in 2024. They emphasised the need to provide training on AI (through the sectoral training funds) in order to improve employees’ knowledge within the sector and futureproof it. These AI-related clauses also refer to Collective Agreement No. 39 of 13 December 1983 on information and consultation regarding the social impact of new technologies. This agreement stipulates that when an employer decides to invest in new technology with significant collective implications for employment, work organisation or working conditions, they are obliged to provide written information on the nature of the new technology, the factors justifying its implementation and its social impact. They must also consult with employee representatives on the social consequences of its introduction. This consultation must cover: employment prospects, workforce structure and planned social measures related to employment, work organisation and working conditions; employee health and safety; and skills development and potential training or retraining measures. These collective bargaining talks referred to the joint declarations of the European social partners in the insurance sector in 2021 and the banking sector in 2024.

In Ireland, the Public Service Agreement 2024–2026(opens in new tab)This link opens in a new tab featured a section called ‘Innovation and digital transformation for service delivery’, under which it was agreed that government and public service unions ‘acknowledge that national competitiveness will depend on the degree to which Ireland can successfully embrace and adapt to developments to do with digitalisation and artificial intelligence (AI)’ and that the parties ‘further agree that the public service must play its part in taking a lead role’. The agreement also provides for ‘cooperation with change and the use of technology to deliver services digitally. This will include maximising the benefits of modern and emerging information technology including Artificial Intelligence and related technologies, Robotic Process Automation and Data Analytics’.

In Denmark, the integration of AI into the workplace became a topic in collective bargaining in 2024 with the Hilfr2 agreement concluded between the cleaning platform Hilfr and the trade union 3F. 3F is Denmark’s largest trade union, with approximately 251,819 skilled and unskilled members as of January 2025. The Hilfr2 agreement explicitly addresses several concerns arising from AI use, including transparency, accountability and workers’ rights. A key element in the agreement is the option to acquire employee status for all workers. Unlike the previous agreement, Hilfr2 ensures that all cleaning assistants are classified as employees rather than independent/self-employed contractors. This classification is pivotal for applying collective bargaining protection to decisions made by AI systems, such as algorithm-driven task assignments or performance evaluations. Under the agreement, Hilfr, as the employer, assumes full responsibility for algorithmic decisions, which can be contested through Denmark's labour dispute resolution system. The agreement expands on EU regulations by requiring Hilfr to take responsibility for explaining and justifying decisions made by algorithms. This includes outlining the basis for decisions and addressing potential procedural harms caused by the use of digital systems. To overcome the challenges of organising a geographically dispersed workforce, the agreement includes a provision for 3F to place a digital link to its website directly on Hilfr’s platform. This link allows workers to access a ‘digital clubhouse’ where they can receive advice, elect representatives and discuss working conditions without employer oversight. By combining collective rights with individual protections, the Hilfr2 agreement sets a precedent for regulating AI in the workplace. It highlights the importance of ensuring workers’ collective rights in a digital labour market, especially in sectors where algorithmic management dominates. Looking ahead, Hilfr2 is expected to inspire future collective bargaining agreements, particularly in platform-based and AI-intensive industries, while continuing to inform international developments, including the EU Platform Work Directive.

In Germany, several collective agreements address aspects of AI, mainly at company level. A 2024 study by the Friedrich Ebert Foundation(opens in new tab)This link opens in a new tab identified eight examples of German collective agreements with AI-related clauses. For example, an agreement in the IBM group(opens in new tab)This link opens in a new tab signed with the group-level works council regulates the conditions for the introduction, operation and expansion of IT systems, including AI. Another example is Deutsche Telekom’s ‘Manifesto for the use of AI(opens in new tab)This link opens in a new tab’, signed by the company and the group-level works council. The manifesto stipulates that the use of AI systems must not have a negative impact on the safety, health and fundamental rights of employees. In terms of the social partners, both the Confederation of German Employers’ Association (BDA(opens in new tab)This link opens in a new tab) and the German Trade Union Confederation (DGB(opens in new tab)This link opens in a new tab) recognise collective agreements between trade unions and employers (federations) and company agreements between works councils and companies as decisive tools in shaping the AI transformation of production and work in favour of workers and companies.

In Spain, examples of collective bargaining on AI are reported at cross-industry, sector and company level. AI in the workplace was one of the key issues in the Spanish cross-industry collective agreement (V AENC). The major trade unions and employer organisations emphasised the importance of human oversight, transparency and fairness. They agreed that companies must provide clear and understandable information to employee representatives about AI-based processes used in hiring, evaluation, promotion and termination. With the overarching goal of preventing bias and discrimination in AI applications, and ensuring the enforcement of workers’ fundamental rights, both parties advocated for collective bargaining to regulate the introduction of AI in the workplace. Additionally, the social partners issued recommendations for negotiating AI clauses(opens in new tab)This link opens in a new tab in collective bargaining(opens in new tab)This link opens in a new tab.

Subsequently, specific provisions on AI began to be incorporated into collective bargaining agreements in Spain in recent years, albeit in an incipient manner. In 2023, according to official statistics, 110 out of 1,773 collective agreements (covering 505,747 employees) regulated the implementation of new technologies. However, the number of specific clauses regulating AI(opens in new tab)This link opens in a new tab may be even lower, according to some experts. As of February 2025, sectoral agreements in banking, credit institutions, insurance, perfumery, private insurance mediation and the food trade in A Coruña, and some company-level agreements (e.g. in El Norte de Castilla, Air Nostrum, Acciona Mobility and AXA) have introduced clauses regulating AI implementation. These clauses focus primarily on protecting workers against indiscriminate use of algorithms and on ensuring transparency regarding AI-driven decisions that impact employment conditions. Some agreements also include job security guarantees and training provisions to help workers adapt to AI technologies.

In France, an article(opens in new tab)This link opens in a new tab published by the National Conservatory of Arts and Crafts (CNAM) in November 2024 analysed the topic of collective bargaining on AI in the information and communications sector, in the finance and insurance sectors and in manufacturing. It found that AI was mentioned in 331 paragraphs in a total of 242 company agreements signed by 160 organisations. Between 2018 and 2023, the proportion of company agreements that mentioned AI doubled. Almost 25% of these agreements were concluded in the information and communications sector. Various sectors adopt different approaches towards AI. In the information and communications sector, the discussion is focused more on training employees and on competition; in the finance and insurance sectors, which are also heavily influenced by AI, worker substitution is the main focus, with terms like ‘automation’ being used. The finance and insurance sectors are becoming more involved in AI discussions, as AI is perceived as a more immediate threat to jobs in companies in these sectors. It is worth mentioning that the most innovative uses of AI in the workplace are to be found in the manufacturing industry, which is moving away from a substitutive logic to focus on workers’ needs, such as training. There is greater discussion of AI in the manufacturing and transport sectors, albeit with a less ‘forward-looking’ approach: AI is not mentioned as a future upheaval, but as an extension of an automation process that is already established. Finally, AI-related mentions are particularly prominent in negotiations about employment: 34.7% of agreements concluded between 2017 and 2024 that address AI mention employment, compared with only 3.9% of all company agreements concluded during the same period. This highlights concerns about job preservation in the context of the digital transformation(opens in new tab)This link opens in a new tab.

Research literature in Norway(opens in new tab)This link opens in a new tab focuses on expectations for AI in the workplace rather than actual experiences and effects of AI. Interviewees in a study conducted among members of the Confederation of Unions for Professionals (Unio) in Norway who use generative AI in their work mentioned that AI tools are useful for brainstorming, sparring, translations, proofreading and producing summaries and meeting notes. Given that AI use is still limited among the professional groups that were mapped (librarians, nurses, police officers, researchers, teachers, radiographers), discussion of the benefits largely revolved around expectations of AI. Findings from the study show that many look forward to AI taking over the ‘boring’ tasks, while others anticipate positive changes in the way they work – without being able to specify exactly what those changes will look like. Among the disadvantages seen by the informants were fears that sensitive data may be compromised, the risk of systemic bias, discrimination and false information generated by AI.

In Czechia, OZ KOVO – covering more than 200,000 employees in more than 400 companies – participated in a regular meeting of the Commission for Collective Bargaining and Social Affairs at industriAll Europe in October 2024, where AI was one of the topics discussed.

In Luxemburg, trade unions have suggested to the employer organisation UEL that a cross-industry agreement on AI should be concluded. The OGBL(opens in new tab)This link opens in a new tab believes, however, that employers are not keen to negotiate an interprofessional agreement but instead want to push the idea of decentralising collective bargaining from the national and sector-level to the company-level.

In Greece, two employer organisations (SEV and GSEVEE) reported initiatives in 2024. SEV (the Hellenic Federation of Enterprises) published the AI Guide, which offers practical insights into AI’s impact on production and management, highlights its potential benefits, and outlines effective strategies for integrating AI into daily business operations. The AI Guide was developed by SEV’s Digital Economy Committee, with special contributions from Ericsson, Huawei, PotamitisVekris, Siemens, Space Hellas, Quest Group and OTE Group. SEV also launched the R U AI? initiative in 2024. Its aim is to help Greek businesses understand the practical benefits of AI in a direct, accessible manner, and to guide them into taking the first steps in training their workforce. This initiative includes the R U AI? Talks video series, in which prominent scientists and professionals share their expertise and perspectives on AI-related topics.

In 2024, GSEVEE (the Hellenic Confederation of Professionals, Craftsmen and Merchants) published a report entitled Greek small and medium-sized enterprises in the era of artificial intelligence. It examines the evolving relationship between AI and SMEs; the challenges and opportunities presented by AI; the key principles of AI and the tools that Greek SMEs can adopt; ongoing efforts to regulate and integrate AI at both national and European level; and early attempts to include SMEs in the relevant strategic planning.

In Slovenia, GZS (the Chamber of Commerce and Industry of Slovenia) established the AI for Slovenia initiative (AI4SI)(opens in new tab)This link opens in a new tab; the chamber is a member of the European AI Forum. The initiative brings together Slovenian providers of AI solutions and services. It aims to accelerate the translation of AI into practice by helping users gain knowledge of and insight into best practices. It also wishes to become a partner in developing strategic plans and legislation in this field. The president of AI4SI explained that its core mission is to promote best practices in the implementation of AI in companies through training, expert events, manuals and other forms of cooperation.

Many countries report that the challenges posed by AI are not directly addressed in collective labour agreements. However, some collective agreements address, in more detail than before, wider digitalisation topics. These include the use of teleworking, such as the collective labour agreement negotiated at the level of the collective bargaining sector – banking activities(opens in new tab)This link opens in a new tab in Romania, for example (2024–2026). This also includes an outline of the right to disconnect.

This clearly shows that national developments are inspired and facilitated by European developments, for example as the European Social Partner Framework Agreement on Digitalisation(opens in new tab)This link opens in a new tab of June 2020. This agreement mentions AI and modalities of connecting and disconnecting. The link between national collective bargaining in the banking and insurance sectors at national level and the joint European social partner declaration in those sectors is another example of this articulation. A European Commission Liaison Forum meeting in early 2025, gathering all European social partner organisations from all sectoral social dialogue committees, focused on the topic of AI. At multinational company level, more and more European works councils are also developing initiatives to provide a framework for the use of AI and its impact on working conditions. Given these examples of collective bargaining and social dialogue on AI in 2024, it can be concluded that AI is an increasingly important phenomenon in social dialogue and an emerging new field of industrial relations research.