Policymaker Toolkit

Proposals to address gaps in existing protections

As algorithmic management systems are becoming embedded in workplaces around the world, so too are the harms which span from algorithmic opacity to mental and physical harm. Policymakers around the world have become aware of the need to tackle the challenges that arise from the development and deployment of algorithmic management systems. In Europe, existing and newly established laws including the General Data Protection Regulation (GDPR), EU AI Act, and Platform Workers Directive (PWD) recognise the harms that algorithmic management poses to employees. While some of these harms can be addressed by existing protections, these frameworks can be patchy and still in a nascent stage. 

In our toolkit for policymakers, we draw attention to some of the gaps and shortcomings existing in the current laws and regulations which do not adequately address algorithmic management. Following this, we offer potential ways forward that policymakers can begin to tackle these challenges centred around improving the measures that already exist. Where possible, for each of normative proposal, our toolkit offers an example of a similar precedent that can inform what improved algorithmic management protections look like.

Our toolkit summarises and extracts key information from over five years of research conducted by the iManage team, led by Professor Jeremias Adams-Prassl. Our analysis focuses on the UK and EU contexts, but may be applicable to other jurisdictions with similar legal landscapes or be drawn upon for comparative analysis. The toolkit only offers a snapshot of our research, with the full published papers linked where applicable and a further research list at the bottom of the page. 

Where are the gaps in existing protections for algorithmic management? How best should they be filled? 

“The solution thus lies in strengthening, complementing, and particularising existing protections, through both legislative and non-legislative measures, both domestically and at the Union level.”  – Jeremias Adams-Prassl, Halefom Abraha, Aislinn Kelly-Lyth, Sangh Rakshita, and Michael ‘Six’ Silberman

Woman working simultaneously as a delivery worker, remote employee, and caring her child, representing the opportunities and risks of work digitalization.
Julieta Longo & Digit / Better Images of AI / CC BY 4.0

Data protection and algorithmic management

While data protection laws in the EU and UK apply to employment contexts in both the private and public sector, these rules were not designed with unique labour relations in mind. As a result, data protection laws suffer from exceptions, structural deficits, fragmentation, and uncertainties which limit their effectiveness in the algorithmic management context. 

Discrimination law and algorithmic management

Doctrinal analysis of the equality framework in the UK and EU by Adams-Prassl et al reveals that its rules are capable of adapting to the harms that may arise from discriminatory and biased algorithmic management systems. Indeed, where issues may arise in terms of application, this largely exposes long-standing unresolved questions in equality law, or stem from the judicial interpretations of discrimination as a human (not algorithmic) phenomenon. In the face of this, reform should be focused on securing effective enforcement of discrimination law as opposed to reforming substantive aspects of discrimination laws. 

Labour law and algorithmic management

The Platform Work Directive (PWD) has been enacted to improve working conditions in the platform economy and regulate algorithmic management across the EU. It provides a critical first response to protect platform workers who were acutely exposed to the harms of automation. Yet, following the enactment of the PWD, a curious paradox has emerged since platform workers, long unprotected by the law, are likely to have more protections against intrusive monitoring and error-prone algorithmic management than 'traditional employees'. Furthermore, the beneficial protections of the PWD depend on faithful and good-intentioned interpretation of the provisions by platforms to meet legislative goals (see Article 1, PWD). 

AI law and algorithmic management

Many algorithmic management systems fall within the scope of the definition of "AI" (Article 3(1), EU AI Act). Indeed, these technologies are also likely to fall within the scope of 'high-risk AI' and thus will be subject to a specified set of requirements which fall on the AI provider (the person/body who develops or markets the AI) (Article 6, EU AI Act; Article 16, EU AI Act; Annex III, EU AI Act). The AI Act is a welcome development which requires algorithmic management providers to take clear risk mitigation measures before deploying high-risk AI systems, however, its effectiveness is diluted by several key structural and practical failures in how it regulates algorithmic management (Kelly-Lyth, 2021).

Looking for a set of regulatory options for algorithmic management?

An outstretched arm holding a print version of the policy blueprint booklet.

Our 'Regulating AI at Work: Policy Blueprint' sets out eight regulatory options to address the regulatory gaps exposed by algorithmic management. The blueprint goes beyond providing comprehensive normative solutions. It also presents concrete illustrations from jurisdictions across the EU and beyond of how these have been implemented in practice.

 

Explore our policy blueprint for regulating AI at work here

This site is part of a project that has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (Grant agreement No. 947806). 

If you have any questions or feedback about this site or our research, please get in touch by emailing ai.work@law.ox.ac.uk. 

Logos for EU emblem, ERC logo, University of Oxford and Bonavero Institute of Human Rights

On this page

Our algorithmic management toolkit for policymakers informs government and regulators about the gaps present in existing protections which apply to algorithmic management. It then offers normative solutions to best fill these gaps.