ILO study highlights gaps between AI ethics and labour standards as workplace AI expands

The International Labour Organization (ILO) has called for closer alignment between artificial intelligence (AI) ethics frameworks and international labour standards, warning that issues such as data privacy, algorithmic transparency and AI-driven workplace management require greater attention as technology reshapes the world of work.

In a working paper titled AI Ethics and the ILO’s international labour standards system: exploring synergies, ILO researchers Jordi Agusti-Panareda and Daniel Samaan examine how five widely recognised AI ethics principles relate to the organisation’s established framework for protecting workers’ rights and promoting social justice.

The paper identifies significant areas of convergence between the two frameworks, particularly around justice, human well-being and the prevention of harm. However, it also highlights gaps in how mainstream AI ethics approaches address collective worker representation, privacy and the governance of AI-supported employment decisions.

The authors argue that the ILO’s international labour standards provide a well-established reference point for governing AI in employment, while AI ethics principles can help inform how existing labour protections are interpreted and applied in technology-driven workplaces.

AI ethics guidelines often overlook labour standards

The paper notes that AI ethics guidelines have proliferated across governments, businesses, international organisations and academic institutions. However, their connection with established labour standards remains limited.

Citing a review of 245 AI ethics documents, the authors note that more than 60 per cent discuss AI ethics in the context of the world of work, but fewer than 20 per cent mention the ILO and its international labour standards.

The researchers identify five widely used principles of AI ethics: beneficence, non-maleficence, autonomy, justice and explicability. These cover the promotion of human well-being, the prevention of harm, human decision-making authority, fairness and the ability to understand and hold AI systems accountable for their decisions.

While these principles provide broad guidance on responsible AI development and use, the paper argues that they often lack the specific mechanisms needed to address workplace concerns.

Unlike voluntary AI ethics guidelines, the ILO’s normative framework is grounded in its Constitution, international labour Conventions and Recommendations, and supervisory practices developed over more than a century.

The authors argue that this framework can help translate broad ethical commitments into more concrete protections for workers and employers as AI adoption grows.

Privacy and algorithmic decision-making emerge as key concerns

One of the paper’s central findings is that AI ethics can help identify emerging risks that may require greater attention when applying international labour standards.

These include the collection and use of personal data, workplace surveillance, algorithmic transparency, digital autonomy and psychosocial risks associated with AI-enabled management.

AI systems are increasingly used in recruitment, performance monitoring and other employment decisions. When these systems rely on opaque scoring mechanisms or inferred personal characteristics, workers may have limited opportunities to understand, challenge or correct the information influencing decisions about them.

The authors highlight the importance of being able to explain AI-supported decisions and identify potential bias, particularly when systems influence hiring, performance evaluations or dismissal.

They argue that transparency is essential to ensuring that AI is not used to conceal practices that conflict with fundamental principles and rights at work.

The paper also points to the potential for AI-driven data collection to shift the balance of power towards organisations that control employee information. Greater access to workers’ behavioural data and inferred characteristics could affect privacy, individual autonomy and the ability of workers to negotiate employment conditions.

Although the ILO framework addresses workplace safety and health, including psychosocial risks, the authors say privacy has historically received less emphasis within its fundamental principles and rights framework than it has in AI ethics debates.

They suggest that these emerging concerns warrant further consideration in the interpretation and application of labour standards.

Collective worker representation remains a gap in AI ethics

The paper identifies another important difference between AI ethics and the ILO’s normative framework: their treatment of autonomy.

Mainstream AI ethics approaches tend to focus on individual autonomy, including the extent to which people retain control over decisions delegated to AI systems.

The ILO framework also recognises individual freedom but places greater emphasis on collective autonomy through freedom of association, collective bargaining and worker representation.

These mechanisms allow workers and employers to organise, represent their interests and participate in determining employment conditions.

The authors argue that collective bargaining has no clear equivalent among the five AI ethics principles examined in the paper. Its limited representation in mainstream AI ethics frameworks highlights a gap in how collective decision-making and the institutional dimensions of workplace relations are considered in AI governance.

As AI becomes more embedded in managerial decisions and employment processes, the paper suggests that governance approaches must consider not only whether humans retain oversight of technology, but also which groups have a voice in decisions affecting their working lives.

ILO standards can guide workplace AI governance

The researchers argue that the ILO’s established rights-based framework can provide a foundation for assessing the opportunities and risks associated with AI in employment.

Its principles, including non-discrimination, freedom of association and the right to collective bargaining, offer benchmarks for evaluating how AI systems affect workers.

The paper also highlights the potential role of AI ethics principles in strengthening the application of existing standards. For example, the principles of justice and explicability could help identify discriminatory outcomes in AI-assisted recruitment and ensure that decisions can be examined and challenged.

The authors cite an example involving an AI-assisted recruitment platform that allegedly used information from CVs, employer-provided data and third-party online sources to generate candidate scores and rankings. Applicants reportedly had limited opportunities to understand the information used, correct it or challenge adverse outcomes.

The case remained pending at the time of writing, and the paper does not take a position on its merits. However, the example illustrates how AI-assisted hiring can raise questions about privacy, transparency and equal opportunity.

The authors point to the ILO’s Discrimination (Employment and Occupation) Convention, 1958 (No. 111), as a relevant benchmark for assessing whether AI-supported recruitment practices undermine equality of opportunity and treatment in employment.

They also suggest that the ILO could examine AI’s effects across different economic sectors and consider whether further guidance or standard-setting measures are needed.

The paper stresses that AI ethics should complement, rather than replace, the obligations arising from international labour standards.

Further research needed as AI reshapes work

The authors conclude that AI ethics and international labour standards should be developed in closer dialogue as AI continues to transform workplaces.

While both frameworks share commitments to human dignity, fairness and well-being, the ILO’s framework offers a more developed legal and institutional foundation for addressing employment-related issues. AI ethics, meanwhile, draws attention to emerging challenges involving data governance, surveillance, algorithmic decision-making and individual digital autonomy.

The researchers call for further work on how privacy, transparency and other AI-related concerns should be reflected in labour-specific rules and practices.

They also emphasise the importance of the ILO’s tripartite governance structure, which brings together governments, employers and workers, in shaping discussions on AI and the future of work.

The paper argues that integrating AI ethics with established labour protections could help ensure that technological progress supports decent work and social justice, while safeguarding workers’ rights in increasingly AI-enabled workplaces.