A N A LYS I S Hilde Wisløff Nagell December 2025 Gender& AI at Work Strengthening OSH to Address Algorithmic Risks FES Nordic Countries

Imprint Published by Friedrich-Ebert-Stiftung e.V. Godesberger Allee 149 53175 Bonn, Germany info@fes.de Issuing Department FES Nordic Countries Barnhusgatan 10 111 23 Stockholm, Sweden Responsible for Content Kristina Birke Daniels| FES Nordic Countries https://nordics.fes.de Responsible Project Manager Meike Büscher| FES Nordic Countries meike.buscher@fes.de Copyediting Carrie Hampel Design/Layout pertext| corporate publishing www.pertext.de The views expressed in this publication are not necessarily those of the Friedrich-Ebert-Stiftung(FES). Commercial use of media published by the FES is not permitted without the written consent of the FES. Publications by the FES may not be used for electioneering purposes. December 2025 © Friedrich-Ebert-Stiftung e.V. Further publications of the Friedrich-Ebert-Stiftung can be found here: ↗ www.fes.de/publikationen

Hilde Wisløff Nagell December 2025 Gender& AI at Work Strengthening OSH to Address Algorithmic Risks

Contents Summary ........................................................  3 Introduction and Background ........................................  3 A Gender Perspective on Occupational Safety and Health(OSH) ...........  4 Monitoring AI Systems for Gender Bias: A Norwegian Example ............  5 The AI Literacy Gap ................................................  6 Trade Unions and Capacities to Act ...................................  7 Conclusion .......................................................  7 Bibliography ......................................................  8 References .......................................................  8

Summary In light of gender-specific organisational safety and health(OSH) risks associated with algorithmic and artificial intelligence systems in the workplace, this paper proposes measures to mitigate discrimination risks. A“builtin” approach to the monitoring of AI-systems is purposed, inspired by Norwegian guidelines for non-discriminatory AI(Vik 2023). It is further argued that the AI-literacy gap needs to be bridged, and that unions have a central role in requiring transparency and co-determination when new AI-systems are introduced. This paper draws on issues and solutions highlighted by international experts taking part in discussions from a workshop – AI-Driven Work: Implications for Women’s Safety, Health, and Equality hosted by FES Future of Work, FES Nordic Countries and the ETUC. The workshop was part of the conference – Towards Gender-Inclusive Digital Policy: AI and Gender Equality in the Workplace that took place in Brussels on the 11th of June 2025. Introduction and Background The integration of algorithmic and artificial Intelligence systems across sectors offers significant opportunities for innovation but also new risks that should be accounted for. AI and algorithmic systems introduce gender-specific risks that can compromise workers’ safety and well-being. These risks are evident in hiring, workplace surveillance, performance evaluation, exposure to harassment, and access to opportunities. If ignored, such risks can result in harmful outcomes such as stress, discrimination, and unfair treatment, and can negatively impact performance, employee turnover, skill-levels and innovation. This paper provides a framework for addressing risks from a gendered perspective in occupational safety and health(OSH) when regulating algorithmic and AI systems in the workplace. Artificial intelligence(AI) denotes machine learning systems that use data and computational models to perform tasks such as prediction, classification, recommendation, and decision making in ways that resemble or complement human cognitive functions, thereby influencing physical or digital work processes. While economics has driven considerable focus on the way AI affects employment, much less attention has been paid to its role in management and human resources. This is partly because analyses most commonly rely on models that treat jobs as bundles of tasks with varying susceptibility to automation. Such approaches highlight the content of work but overlook the context of work—particularly the ways workers are managed and evaluated(Owen 2024). In the context of work, AI does present potential benefits: job-matching systems guide workers toward better opportunities; algorithmic hiring tools have the capacity to reduce some forms of human bias; and for certain workers, being evaluated by an algorithm may feel less arbitrary than facing the whims of a difficult manager. However, AI-driven workplace management tools may also risk reducing worker autonomy, heighten surveillance, and create feelings of isolation(OECD 2022). They can contribute to increased employee turnover, deskilling, stress, and other health-related consequences. Many of these effects carry a gendered dimension, reinforcing discrimination, undermining well-being, and exacerbating inequality(UN Women 2024). When hardware and software systems are not designed with diverse bodies, working processes and needs in mind, particularly the needs of women, they risk reinforcing stereotypes, misjudging capabilities, and fostering unsafe or exclusionary environments. In much the same vain, robotics that overlook ergonomic diversity can physically disadvantage some workers, either through discomfort or by excluding them from specific tasks. In addition to the risk of deficits in accommodating the needs of all workers, emerging evidence is showing that when women and men use AI to produce identical work, women are perceived as less competent than men. During discussions at the FES workshop, AI-Driven Work: Implications for Women’s Safety, Health, and Equality, participants referred to the known biases of AI training data, and the problematic and growing use of algorithm-driven systems in personnel management, with Microsoft Viva Insights as a notable example. These systems track a wide range of employee data— such as emails sent, phone calls made, days of absence, and log-in times—and aggregate it into actionable insights on behaviour, engagement, and well-being. The insights are then presented to team leads in intuitive charts and graphs, enabling management to monitor and influence employee performance. This is often framed as a positive development, as decisions by team leads or management are now supposedly supported by neutral data, and less influenced by human bias. This is obviously not the case, as the systems incorporate biases via training data. One of the discussants at the workshop, Miriam Klöpper, has elaborated on these points in her work on the subject, where she also stresses that while algorithms may appear neutral, they cannot solve deeper problems in isolation(Klöpper 2025). Advancing fairness at work re lies on strong social frameworks and on empowering individuals to challenge unjust decisions. Ensuring that everyone, regardless of gender, is able and encouraged to do so is fundamental for real change. AI-driven surveillance and performance tools, such as algorithmic management systems, are increasingly used in workplaces for tasks like hiring, tracking productivity, Introduction and Background 3

and monitoring emotions. These tools also claim to enhance workers wellbeing. While the aim is to enhance efficiency, they can unintentionally cause harm, amplify workplace inequalities, and may discriminate women due to biased data. This can occur during recruitment, daily tasks, and emotional assessments, leading to unfair judgments and reinforcing biases. AI and robotics technologies do not simply support work; they shape it. They structure workplace conditions, influence employee behaviour, and directly affect both physical and mental health. Identifying OSH risks can help build workplaces that are healthier, more equitable, and more sustainable for all, as well as ensuring a better environment for innovation. The use of AI requires the implementation of proactive policies that address the new OSH-risks associated with digitalisation and automation. A Gender Perspective on Occupational Safety and Health(OSH) In recent years, there has been growing—but still limited—attention from international organisations, research institutes, and policymakers to the gendered implications of AI in employment and workplace dynamics. Research highlights the persistent underrepresentation of women in AI professions, the prevalence of algorithmic bias in hiring and promotion, and the critical importance of gender-sensitive data for equitable AI deployment (FES Future of Work 2025). AI-driven automation can both reinforce and mitigate inequalities, depending on how data and design decisions are handled; exclusion from data can entrench discriminatory outcomes, while inclusive practices can help close the gap. However, left unchecked, gender bias in AI can reinforce stereotypes and create hazardous or unfair workplaces, affecting everything from job assignments to promotion decisions and disciplinary actions and severely affect workers’ safety and wellbeing. A gendered perspective in occupational safety and health(OSH) is essential because algorithmic management creates risks that affect men and women differently. Traditional OSH frameworks focus on physical hazards or general psychosocial risks but often overlook how biased algorithms, surveillance systems, and digital tools can exacerbate discrimination, harassment, and exclusion in the workplace. Without integrating gender, OSH policies fail to address structural and indirect harm, from biased performance monitoring to online harassment. Updating OSH to include gendered risks ensures safer, more equitable workplaces and equips organisations to mitigate both physical and psychosocial harm caused by emerging technologies. Legislation addressing emerging technologies already exists at the national level and through national applications of EU law(Eurofound 2024), such as AI and algo rithmic management, and collective bargaining in relation to these technologies, while some collective agreements in Europe now explicitly include provisions on the issue(UNI Europa 2024). At the same time, un ions, experts, and political actors are calling for a stronger reaffirmation of collective bargaining rights at the EU level. The AI Act Regulation(2024) established a legal framework obliging developers and AI system users to address bias, increase transparency, and ensure accountability. For a closer examination of the AI Act and its capacity to address gender inequities, see Karagianni 2025 (Karagianni 2025). The EU AI-act is a step in the right direction. The firstin-the world law on artificial intelligence may help identify gender bias and ensure that workers’ safety and well-being is not negatively affected However, rather than passively wait for regulations to apply, affected 1 parties, employers, employees, organisations and unions can and should take steps to reduce gender-specific risks of AI. A report from the international Labor Organization(ILO) points out regulatory gaps in the managing of OSH-risks related to automation(ILO 2025). Broad conventions pro vide a foundation for the right to a safe and healthy workplace, but there is a lack of more concrete regulations that specify how to secure this right in the workplace AI-systems are never neutral but depend on various choices, from how the model is trained, to how data is labelled by people, and which technologies are chosen, from testing to deployment. As Johannes Anttila, policy advisor in the European Par liament, has mentioned in his work, and during the conference workshop, these systems are“sociotechnical”: They involve people, culture and context. It should not come as a surprise that AI-systems will tend to reflect existing structural inequalities and bias related to gender. Anttila has argued that what we need to do is bring back context, transparency, and worker agency, all the while making sure that fundamental rights are protected. Transparency is one step in this, but what is needed is reflexivity, a purpose and intent in the design and the deployment phases. 1  There is a concern that the Digital Omnibus Package will undermine the protections the Act initiated as protecting privacy, discrimination, transparency, fairness and workers’ rights, which could adversely affect women and other vulnerable groups(Del Castillo 2025). 4 Friedrich-Ebert-Stiftung e.V.

In the following, I will provide an example of how AI ­systems can be regularly monitored for gender bias by following a structured, ongoing process. Monitoring AI Systems for Gender Bias: A Norwegian Example In November 2023, The Norwegian Equality and Anti-­ Discrimination Ombud(Likestillings- og diskrimineringsombudet, LDO) launched a guide in order to uncover and prevent discrimination in the development and use of artificial intelligence. Discrimination is defined as unlawful differential treatment based on protected grounds such as gender, ethnicity, disability, religion, sexual orientation, age, and combinations thereof(compound discrimination). From the perspective of law, it is important to note that the prohibition of discrimination – for instance as it is formulated in Norwegian law – applies regardless of whether decisions are made by humans or AI systems. The guide aims to make stakeholders aware of anti-discrimination legislation and to help them systematically prevent discrimination throughout all phases of AI system development and use. Alex Moltzau at the European AI-office has provided an informative and accessible presentation of the guidelines(Moltzau 2023). The approach in the guide is that of“built-in protection against discrimination”. Measures to prevent discrimination and promote equality must be built into all development phases of an AI system, from planning to use of the technology. Guides with a similar perspective have been developed elsewhere in policy briefs, such as from the Netherlands government called, Fundamental Rights and Algorithm Impact Assessment(FRAIA)(Government of the Netherlands 2021), and from the Finnish government:“Promoting equal ity in the use of Artificial Intelligence— an assessment framework for non-discriminatory AI”(Ojanen et al 2022). “Built-in protection” implies that the AI-technology in question must be monitored from planning to implementation. To avoid discriminatory practice later, critical questions need to be asked in the very first planning phase. At the outset, it is crucial to design the project with an awareness of the potential risks of discrimination. Are the selected data sources and methods representative of the population they will affect? Is there potentially bias in historical data or systemic structures that need to be analysed carefully? What is the anticipated impact of the system on different demographic groups? The planning phase is about laying a fair foundation before data or algorithms are introduced. In the second phase, involving the training of data, the focus shifts to the choice and justification of data variables. The Norwegian guide stresses that variables must be aligned with the defined purpose of the system, and the inclusion of sensitive variables should be explained and defended based on necessity. Comparing different data models and datasets to evaluate how they influence outcomes is also recommended. This step ensures that the model is not reproducing unfair or unjustified distinctions between groups. The third phase is the development of the AI-model itself. During development, clarity of purpose and context is essential. The model’s intent must be well defined, along with identifying which groups could be affected by its decisions or predictions. Involving a broad set of stakeholders—including technical experts, policymakers, domain specialists, and representatives of affected communities—helps to identify risks and keeps the system accountable. This prevents development from being a purely technical process detached from societal concerns. Before deployment, models should undergo rigorous testing to evaluate fairness across different groups. This means testing both for direct discrimination(e.g., a model treating two groups differently due to explicit variables) and indirect or compound discrimination(e.g., when neutral variables unintentionally act as proxies for protected characteristics). An example of direct discrimination could be a hiring model that takes“gender” as an input and systematically assigns lower suitability scores to women than to men with identical CVs and test results. Indirect discrimination is more subtle, like a hiring system that does not see“gender” but uses features such as participation in certain sports, membership in specific fraternities, or patterns in previous job titles that in the historical data are much more common among men. The model then prefers“male coded” CVs and disfavours“female coded” career paths, even though gender is never explicitly provided. These tests help uncover patterns that may lead to systematic disadvantages. Implementation must be followed by monitoring. Once in operation, AI systems must remain under oversight and corrective governance. If discriminatory patterns or biases are detected, corrective measures should be applied, whether through technical adjustments, rule changes, or systemic oversight mechanisms. Importantly, affected individuals must be informed about the system’s function and their rights in relation to its decisions. Building trust requires transparency, accountability, and ongoing monitoring long after the model’s initial release. The Norwegian guide is just one example of systematic monitoring, that can be used as a source of inspiration. What the“built-in design” requires is that questions of non-discrimination are addressed early enough to have an Monitoring AI Systems for Gender Bias 5

Five phases from planning to implementation 5 Implementation and supervision Investigate whether biases in the system can be levelled out by compensatory 5 1 measures. Human control of the system must be ensured, and knowledge of the system made known to those affected, so that they can protect their rights. Figure 1 1 Planning Define formal and intended use of the AI system. The context in which the system is to operate is decisive for what significance it may have on an individual and societal level. 4 Testing Map whether the system works equally precisely for all groups. Even if personal data is removed, the system can find patterns in the data so that information that coincides with the grounds for discrimination is illegally emphasised. 2 4 3 3 Model development The system’s calculation must coincide with the overall formula. The variables the system’s calculations are based on must be relevant. Pavis factual formality, necessity and proportionality for variables that can be linked to a basis for discrimination. 2 Training data The data base model is trained to be representative based on the model’s functionality, and historical biases must be mapped and corrected. Source: New Norwegian Guide to Prevent AI Discrimination Launched With Minister of Digitalisation| by Alex Moltzau| Ethical AI Resources| Medium(Moltzau 2023) impact on important choices determining the way the AI-system is constructed, what data is used, and what implications the system may have for affected groups. Unfortunately, what too often happens is that systems are bought ready-made, and critical questions not asked before the system is already in operation, and sometimes not even then. This may be partly due to lack of knowledge and competence at all levels in the demand-supply chain. The AI Literacy Gap The EU AI Act obliges providers and deployers of AI systems to ensure that providers and deployers of AI systems have a sufficient level of technical knowledge, experience, education and training. AI literacy is defined as the“skills, knowledge, and understanding” needed to deploy AI systems responsibly(Gulley 2025). Research shows women tend to adopt AI tools at dramatically lower rates than men, often due to concerns about ethics, judgment within the workplace, and a perceived lack of expertise. Ensuring that digital literacy education is inclusive and responsive to such gender dynamics is essential for raising awareness and empowering all groups(Blending 2025). A study involving 1,026 software engineers found that when women and men use AI to produce identical work, women are perceived as less competent than men. On average, engineers believed to have used AI were rated 9% less competent, but women who used AI faced a 13% reduction in perceived competence compared to a 6% reduction for men. The“competence penalty” against women using AI is particularly harsh when evaluated by male engineers who themselves do not use AI—they rate female AI users 26% more harshly than male AI users for the same output(Travis 2025). Despite company efforts to encourage AI adoption in software development, only 31% of female engineers used the AI tool versus 41% of all engineers, likely due to concerns about this competence bias. Baseline knowledge and confidence levels varies considerably and there is a gender gap that needs to be bridged by building AI-literacy among women(Gulley 2025). Research also indicates that the differences in low-cost and premium options for AI are furthering inequality among white collar workers. Based on a survey of 4,000 knowledge workers across the UK, US, Germany, and Canada, the study reveals that higher earners have disproportionate access to the latest AI tools and training, 6 Friedrich-Ebert-Stiftung e.V.

allowing them to reap AI’s promised rewards. In contrast, lower earners and women are being shut out from AI opportunities, which impacts their skill development, job satisfaction, and time savings, both personally and professionally(Human Rights Press Releases 2025). Emerging evidence suggests that there is a considerable AI-literacy gap that needs to be bridged. Inclusive upskilling initiatives are essential to give women equal opportunities to build competencies in AI use and to close adoption gaps. Trade Unions and Capacities to Act More can be done at the workplace to protect workers from bias, exclusion, and harassment associated with the introduction of AI-system. A start is to acknowledge that bias in data and algorithms is a major gender-specific risk. AI systems trained on historical datasets may perpetuate discriminatory hiring practices or amplify stereotypes, leading to unfair evaluations or exclusion in recruitment and performance assessments. For example, algorithmic scheduling—common in female-dominated sectors like retail and hospitality—allows shifts to be allocated with little notice, increasing work and income uncertainty and stress for women workers. Advocacy groups, NGOs, but in particular trade unions, play a vital role in spreading knowledge about these specific risks. Workers themselves should also be able to discern when, for example, algorithmic tools used at the workplace are having gendered effects and be aware of their overall right for redress. Transparent reporting requires employers to explain AI decision processes, publish the results of audits, and maintain open channels for complaints. Raising awareness begins with regular transparency measures such as bias audits of AI systems and communicating findings openly to affected employees. Organisations should be encouraged to publish easy-to-understand reports on the risks identified for specific groups and describe how these risks are being addressed so workers become conscious of the underlying issues Furthermore, there are many other ways in which organisations can address the gendered impacts of AI. Workshops and training programs could help highlight potential risks and safeguards, ensuring both workers and managers understand how technology can reinforce or reduce bias. Transparency is a first step, but more importantly, workers must be able to influence and negotiate how algorithms are developed and deployed. Involving unions and women’s groups in the rollout of new technologies ensures that diverse perspectives and worker voices are integrated into the process. Worker’s involvement is a necessary requirement for being able to discern when, for example algorithmic tools used at the workplace are having gendered effects or of their overall rights, for instance of co-determination. It also means involving affected workers in the design and rollout of workplace AI systems. When platforms and algorithms are developed by diverse teams and with meaningful input from women, they are less likely to embed harmful stereotypes. Studies have shown that biases often creep in“by design,” mirroring societal prejudices unless deliberately countered by participatory processes. Workplace committees, user panels, or feedback forums can ensure that the experience and concerns of women are recognised. This not only raises awareness but leads to more inclusive and equitable technologies. Much of this debate focuses on workers’ rights to transparency, participation, negotiation, and redress. Yet, the discussion above also highlights another important aspect: bargaining over algorithms is not only about oversight but also about co-design—ensuring that these systems function as intended. This should be a shared responsibility of both employers and workers, anchored in transparent and well-defined rules(Enriquez/Anttila). Conclusion In this paper, it is argued that AI systems in the workplace must be proactively monitored and regulated to prevent gender bias, protect worker safety and well-being, and ensure fair treatment, especially as AI and robotics become more prevalent. As outlined in the Norwegian guidelines for non-discriminatory AI, monitoring should start at the early phase of planning and development of the AI-tools. Equally important, monitoring must be in place in the implementation phase to protect workers from OSH-risks and actively involve workers and other stakeholders in the process. Increasing awareness of gender-specific AI risks in the workplace demands systemic change—through education, transparency, regulation, stakeholder involvement, and organisational culture. Unions have a particularly important role to play in empowering workers – and especially women – to recognise risks, advocate for improvement, and participate in shaping fair and responsible use of AI. Unions may also help raise critical questions about ownership and control. Who develops these technologies, and what biases are embedded in the data they rely on? Depending on these choices, AI systems can either perpetuate, widen, or help reduce existing gender inequalities. Conclusion 7

Bibliography References FES Future of Work – policy conference, 2025: Towards Gender Inclusive Digital Policy: AI and Gender Equality in the Workplace. with keynote speech from Ivana Bartoletti, vice president, Global Data Privacy and AI Governance Officer at Wipro. The panellists of the Workshop“AI-Driven Work: Implications for Women’s Safety, Health, and Equality” co-hosted by the European Trade Union Confederation(ETUC) and the FES Nordic Countries, were: → Silvia Semenzin, Scholar and Activist, AI Forensics → Johannes Anttila, Adviser, European Parliament → Hilde Nagell, Adviser, Agenda, Norway → Anastasia Karagianni, Researcher, Vrije Universiteit Brussels(VUB) → Miriam Klöpper, Researcher, NTNU, Norway → Aline Brüser, Gender Equality Advisor, ETUC See full speakers list of the conference: https://futureofwork.fes.de/our-projects/­genderinclusive-digital-policy[Attended on 11 June 2025 in Brussels, Belgium.] FEPS – Foundation for European Progressive Studies conference, 2024: How do algorithms and AI reshape workplaces and the world? Presentations: Algorithmic management practices in regular workplaces: Insights from different sectors, Uma Rani, senior economist, International Labor Organization(ILO); Algorithmic management and workplace digitalisation in Finland, Johannes Antilla, senior expert, Demos Helsinki; Annikka Lemmens, economist and policy analyst, Directorate for Employment, Labor and Social Affairs, Organization for Economic Cooperation and Development(OECD). Panel participants: Juha Antila, head of R&D, The Central Organization of Finnish Trade Unions(SAK); Antti Hakala, director of finance sector and ICT and communications sector, Trade Union Pro; Miapetra Kumpula-Natri, Member of Parliament, SDP Finland; Laura Seppänen, chief scientist, Finnish Institute of Occupational Health; Antti Rinne, general secretary, SAMAK; and Gerard Rinse Oosterwijk, policy analyst, FEPS. See: https://feps-europe.eu/event/howalgorithms-and-ai-­reshape-workplaces-and-the-world/[attended on 17 September 2024 in Helsinki, Finland]. Video: https://www.youtube.com/watch?v=yE4UKhjZw5c&t=1067s[accessed 1 September 2025]. Blending, M.(2025): Women Are Avoiding Using Artificial Intelligence – Can That Hurt Their Careers?. Harvard Business School. Available at: https://www.library.hbs.edu/working-knowledge/women-are-avoiding-using-artificial-intelligence-can-that-hurt-their-careers(last accessed on 10.12.2025). Davis, O.(2024): We Need Better Data on Workplace AI. TechPolicy. Press. Available at: https://techpolicy.press/we-need-better-data-on-workplace-ai/(last accessed on 10.12.2025). Del Castillo, A. P.(2025): The Digital Omnibus: Deregulation Dressed as Innovation. Social Europe. Available at: https://www.socialeurope.eu/the-digital-omnibus-deregulation-dressed-as-innovation(last accessed on 10.12.2025). Enriquez, D.; Anttila, J.,(2025): Tech Policy Press: Lessons from Regulating Automated Man agement: What Automation Teaches Us about Supporting Workers. Available at: https:// www.techpolicy.press/lessons-from-regulating-automated-management-what-automation-teaches-us-about-supporting-workers/(last accessed on 10.12.2025). Eurofound(2024): Regulatory Responses to Algorithmic Management in the EU. Available at: https://www.eurofound.europa.eu/en/publications/all/regulatory-responses-algorithmic-management-eu(last accessed on 10.12.2025). Friedrich-Ebert-Stiftung(FES) Future of Work(2025): Artificial Intelligence and Gender Equality. Available at: https://futureofwork.fes.de/our-projects/artificial-intelligence-and-gender-equality.html(last accessed on 10.12.2025). Government of the Netherlands(2021): Fundamental Rights and Algorithms Impact Assessment(FRAIA). Available at: https://www.government.nl/documents/reports/2021/07/31/impact-assessment-fundamental-rights-and-algorithms(last accessed on 10.12.2025). Gulley, A.(2025): Mind the Gap: AI Literacy Requirements Under the EU AI Act and the Gender Divide. Women in AI. Available at: https://www.womeninai.co/post/mind-the-gap-ai-literacy-requirements-under-the-eu-ai-act-and-the-gender-divide(last accessed on 10.12.2025). Human Rights Press Releases. Adaptavist Group(2025): Artificial Inequality: AI Is Exacerbating Career, Income and Gender Divides.Available at: https://www.humanrightspressreleases. com/article/830601737-artificial-inequality-ai-is-exacerbating-career-income-and-gender-divides-research-from-the-adaptavist-group-reveals(last accessed on 10.12.2025). International Labour Organization(ILO)(2025): AI and digitalization are transforming safety and health at work. Geneva: International Labour Organization. Available at: https://www.ilo. org/resource/news/ai-and-digitalization-are-transforming-safety-and-health-work(last accessed on 10.12.2025) Karagianni, A.(2025): The EU Artificial Intelligence Act through a Gender Lens. Brussels: Friedrich-Ebert-Stiftung, Competence Centre on the Future of Work, Analysis. Available at: https://library.fes.de/pdf-files/bueros/bruessel/21887-20250304.pdf(last accessed on 10.12.2025). Klöpper, M,(2025): When Fairness Feels Different: Gender, Algorithms and the Future of Work. Friedrich-Ebert-Stiftung, Future of Work. Available at: https://futureofwork.fes.de/ news-list/e/when-fairness-feels-different-gender-algorithms-and-the-future-of-work.html (last accessed on 10.12.2025). Moltzau, A(2023): New Norwegian Guide to Prevent AI Discrimination Launched With Minister of Digitalisation. Medium Ethical AI Resources. Available at: https://medium.com/­ ethical-ai-resources/new-norwegian-guide-to-prevent-ai-discrimination-launched-w­ ithminister-of-digitalisation-6806cc9d0b65(last accessed on 10.12.2025). OECD(2022): The Effects of AI on the Working Lives of Women. Available at: https://www. oecd.org/en/publications/the-effects-of-ai-on-the-working-lives-of-women_14e9b92c-en. html(last accessed on 10.12.2025). Ojanen, A.; Björk, A.; Mikkonen J.(2022): An Assessment Framework for Non-discriminatory AI. Demos Helsinki. Available at: https://demoshelsinki.fi/publication/an-assessment-framework-for-non-discriminatory-ai/(last accessed on 10.12.2025). Travis, M.(2025): Women Who Use AI at Work Face a Predictable Competence Penalty. Forbes.Available at: https://www.forbes.com/sites/michelletravis/2025/08/26/women-whouse-ai-at-work-face-a-predictable-competence-penalty/(last accessed on 10.12.2025). UN Women(2024): Explainer: Artificial Intelligence and gender equality. Available at: https://www.unwomen.org/en/articles/explainer/artificial-intelligence-and-gender-equality (last accessed on 01.09.2025). UNI Europa(2024): A Database of AI and Algorithmic Management in Collective Bargaining Agreements. Available at: https://www.uni-europa.org/news/a­-database-of-ai-andalgorithmic-management-in-collective-bargaining-agreements/(last accessed on 10.12.2025). Vik, K. T. A.(2023): Innebygd diskrimineringsvern: En veileder for å avdekke og forebygge diskriminering i utvikling og bruk av kunstig intelligens. Oslo: Likestillings- og diskrimineringsombudet. Available at: https://ldo.no/content/uploads/2024/06/Innebygd-diskrimineringsvern-2023.pdf(last accessed on 10.12.2025) 8 Friedrich-Ebert-Stiftung e.V.

About the author Hilde Nagell has been a member of the consortium group of the Digital Programme at the Foundation for European Progressive Studies(FEPS), she holds a PhD in political philosophy and is currently working as an advisor at the progressive think-tank Agenda in Norway.

Gender& AI at Work In light of gender-specific organisational safety and health(OSH) risks associated with algorithmic and artificial intelligence systems in the workplace, this paper proposes measures to mitigate discrimination risks. A“built-in” approach to the monitoring of AI-systems is purposed, inspired by Norwegian guidelines for non-discriminatory AI. It is further argued that the AI-literacy gap needs to be bridged, and that unions have a central role in requiring transparency and co-determination when new AI-systems are introduced. Further information on the topic can be found here: ↗ fes.de