DOWNLOAD POLICY PAPER
Please cite as:
Pretorius, Christo. (2024). “EU Employment Law and the AI Act: A Policy Brief Putting the Human Back in ‘Human-Centric’ Policy.” Policy Papers. European Center for Populism Studies (ECPS). September 11, 2024.https://doi.org/10.55271/pop0002
This policy paper analyzes the European Union’s (EU) AI Act, aimed at regulating Artificial Intelligence (AI) through four risk classifications related to data protection, privacy, security, and fundamental rights. While the Act establishes regulatory frameworks, it neglects employment security, a critical factor behind public mistrust of AI. The paper warns that failure to address this issue could deepen socio-economic inequalities and lead to political unrest. Recommendations include promoting collective negotiation between workers and employers, advocating for legislation on redundancies linked to AI, and launching information campaigns to educate workers, thus ensuring fair working conditions and improving trust in AI technology.
Executive Summary
The European Union (EU) is attempting to regulate the deployment of Artificial Intelligence (AI) through the recently passed AI Act. Overall, the Act outlines four distinct classifications for AI systems, categorizing them on the risk they pose on an individual’s data protection, privacy, security, and fundamental rights. It further provides regulations and guidance to member states on each category and calls for the establishment of national and EU level regulatory bodies to enforce the Act. However, ultimately the Act overlooks the critical issue of employment security, which is the main cause behind mistrust of AI. This gap could exacerbate socio-economic inequalities and fuel political unrest in the short to long term if it is not addressed promptly.
Research indicates that AI will have a disruptive effect on employment overall as certain types of work is automated and augmented, but the effects of this will be felt most in clerical, secretarial, and para-professional roles, which poses a risk to vulnerable groups including women and those with lower educational attainment. There is a pressing need for proactive measures to mitigate the harmful effects of the technology’s implementation on workers and their families, specifically the protection against unjustified dismissal and the assurance of fair working conditions. The following recommendations are proposed to address the Act’s shortcomings:
Collective Negotiation: Encourage cooperation between workers, employers, and worker associations to assess AI’s impact on jobs. This could lead to agreements on redeployment, education opportunities, or redundancy notices, providing workers with clearer timelines and reducing workplace disruption.
Advocacy for New Legislation: Push for legislation that mandates notice periods for redundancies due to technological innovation, building on the EU Directive on Transparent and Predictable Working Conditions. The International Labor Organization’s 1982 Termination of Employment Convention offers a valuable template for such legislation.
Information Campaigns: Launch campaigns to educate workers on AI systems, their potential benefits, and available upskilling opportunities. These efforts would enhance trust in AI, aligning with the AI Act’s goal of ensuring human-centric, safe, and lawful AI deployment.
Context: The Problem with the AI Act
The discussion about AI regulation in the European Union (EU) began with Ursula von der Leyen’s 2019-2024 agenda for Europe, A Union that Strives for More (2019). It stated that she would ‘put forward legislation for a coordinated European approach on the human and ethical implications of Artificial Intelligence’ (von der Leyen, 2019: 13). What followed was a call for greater focus on enabling more investment in, and the better coordination of the development and deployment of AI in the EU, alongside a call for a clear definition of what high-risk AI systems are (General Secretariat of the Council to Delegations, 2020). Jointly, the Coordinated Plan on Artificial Intelligence was published to help foster trust in AI systems yet failed to address the real consideration that such a disruptive technology would have on the world of work (European Commission, 2021). The purpose of this policy brief is to advocate for greater attention to be given to the area of employment law, so that action may be taken to ease the concerns over job loss relating to AI, and thus, foster greater trust in this new technology.
A report from the International Labor Organization estimated that the introduction of AI systems would have an overall disruptive effect worldwide, highlight that an important share of clerical, secretarial, and para-professional jobs would be most affected (Gmyrek et al., 2023). These findings are supported by similar ones from PricewaterhouseCoopers International Limited (PWC), the US-based National Bureau of Economic Research, and the Pew Research Centre, which also found that women and individuals with lower levels of education are most vulnerable to job loss due to automation (Hawksworth, Berriman, & Goel, 2018; Acemoglu, & Restrepo, 2021; Kochhar, 2023). Although this is an evolving issue, as AI indeed has the potential to increase economic growth and improve lives, the disruptive impact of AI systems on the world of work has yet to be felt. Therefore, it is important to take the necessary steps now to mitigate the harmful affect this technology can have on workers and their families and provide a safety net to individuals during this period of transition.
Currently the EU’s AI Act states that it is trying to ensure ‘a high level of protection of health, safety, fundamental rights as enshrined in the Charter of Fundamental Rights (2012) of the European Union (hereby referred to as the ‘Charter’), including democracy, the rule of law and environmental protection, to protect against the harmful effects of AI systems in the Union…’. However, one of the fundamental rights not addressed was that of Articles 30 (Protection in the event of unjustified dismissal) and 31 (Fair and just working conditions) of the Charter (European Union, 2012). The focus on adopting AI throughout the EU, whilst highlighting the issues of data protection, privacy, and security of individuals in the long term, has left the area of employment concerns unaddressed.
The EU has highlighted that this area is one that should be legislated on in the future, stating in their 2022 report on artificial intelligence in the digital age: ‘[The European Parliament] emphasises that the use of AI in this area gives rise to a number of ethical, legal and employment related challenges… [and] stresses that AI is currently already substituting or complementing humans in a subset of tasks but that it is not yet having detectable significant aggregate labour market consequences’ (European Parliament, 2022). The report also highlights that strong links between AI and the rising socio-economic inequality has been found, and researchers warn that unregulated AI will further increase the wealth gap within society, a finding supported by other academic publications (Rotman, 2022; Bushwick, 2023). MEP Brando Benifei indicates that a proposal for a directive on AI at the workplace is something to be discussed in the future, but at present there is not much information from Benifei or anyone else on what policy discussions in this area will look like (Publyon, 2024). This uncertainty continues following von der Leyen’s announcement that more research investments will be made into AI, leaving commentators to speculate what will actually be funded (Wold, 2024). Much like the events of 2008, people that feel left behind by the increasing automatization and augmentation of the workplace could be persuaded into more extreme populist politics, which is why delaying discussions on this topic are problematic (Steiner et al., 2023). While the EU continues to refine policy, the implementation of AI into the workplace is happening now, and the situation for workers could change rapidly as new technology hits the market. Just as the EU attempted to get ahead of AI by defining it, they should attempt to get ahead of employment concerns before they evolve past being concerns alone.
The AI Act in Brief
To understand the shortcomings of the AI Act (2024), it is helpful to summarize the contents of the regulation first. The European Union’s AI Act came into force on the 13th of June 2024, seeking to create rules that would govern the development, employment, and use of artificial intelligence (AI) systems within the Union. The rules define four levels of risk regarding AI systems, offering only a description of what each category is, and indicate what systems are I place to regulate them within the common market:
Unacceptable Risk: These AI are banned within the EU, except in limited circumstances.
[Article 5.1(a/b)] Manipulative AI that can deceive, subvert, or impair autonomy, decision-making and free choice. This includes AI that may exploit disadvantaged persons whether through socio-economic vulnerabilities or take advantage of a disability. Notable exceptions to this are AI used in the context of medical treatment such as psychological treatment of a mental disease or physical rehabilitation, or within advertisement.
[Article 5.1(c)] AI systems that provide social scoring as it may lead to discrimination and exclusion.
[Article 5.1(d)] Risk assessment or predictive AI systems in the context of law enforcement.
[Article 5.1(e)] AI systems that scrape footage to expand facial recognition databases.
[Article 5.1(g/h)] Biometric categorization systems that are based on natural persons’ biometric data, such as an individual person’s face or fingerprint, to deduce or infer an individuals’ political opinions, trade union membership, religious or philosophical beliefs, race, sex life or sexual orientation. The notable exception to this rule is biometric categorization systems employed by law enforcement agencies for anti-terrorism and missing persons.
High Risk: [Preamble (paragraph 48)] These systems are defined as having a negative effect on safety or the following fundamental rights, the latter of which is in this case a person’s access to education, employment, public or private services, legal representation, and administrative or democratic processes:
| The protection of personal data, | The rights of persons with disabilities, |
| Freedom of expression and information, | Gender equality, |
| Freedom of assembly and of association, | Intellectual property rights, |
| The right to non-discrimination, | Workers’ rights, |
| The right to an effective remedy and a fair trial, | The right to education, |
| Consumer protection, | The right to good administration |
| The right of defense and the presumption of innocence, |
High risk systems are also classified as that related to critical infrastructure and biometric identification systems.
Limited Risk: [Preamble (paragraph 53)] AI systems that do not materially influence the outcome of decision-making, and/or augment tasks that are either automated or conducted by humans are considered to be limited risk. This category is highlighted as needing further guidelines in the future.
Minimal Risk: Most AI systems currently available fall under this category – they provide solutions with minimal risk, and are therefore not regulated, nor will be moving forward.
The regulation achieved most of the mandated aims, creating clear definitions for different levels of risk, whilst focusing on the impact different AI systems could have on individuals.
Recommendations
The need to deal with the issue of job loss and employment law concerns are real and present and must be addressed in a timely manner during these early stages of AI implementation. Although further training opportunities is an avenue that the EU is pursuing, to contribute to the Commission’s call for the development of an ecosystem of trust by proposing a legal framework for trustworthy AI, this paper proposes three different avenues the issue of employment security could be addressed:
Collective Negotiation: Close cooperation between worker associations, the groups represented by them, and employers, can allow for investigation on the potential disruptive impact that AI systems can have on various professions. This will allow them to make informed decisions so that they may take steps to reach collective agreement that would allow for either redeployment of workers, advertise or make available further education opportunities, or have a guaranteed period of redundancy notice with regards to the implementation of AI systems. Similarly, management could make available to workers a clear AI implementation plan so that workplace disruption is reduced, and workers know in advance the time they have should their work be made redundant.
Advocacy For the Adaption/Adopting of New Legislation: If no provisions are currently in place, countries within the EU must take active steps to create or adapt legislation that will allow workers to have a notice period that they will be made redundant due to technological innovation. The EU Directive on Transparent and Predictable Working Conditions in the European Union, 2019, set a precedent that the EU was willing to use its competency in the area of social rights to address new forms of employment to protect EU workers from unpredictable employment (Official Journal of the European Union, 2019).
Given the close relationship between the International Labor Organization (ILO) and the EU, the ILO’s 1982 -Termination of Employment Convention (C158) would provide a good basis for text that could be incorporated into the EU Directive on Transparent and Predictable Working Conditions in the European Union (Pretorius, 2023). Only nine nations have ratified the convention, which contains the following article that could either be incorporated into EU law, or provide an example that can be followed:
‘[Article 13.1.] When the employer contemplates terminations for reasons of an economic, technological, structural or similar nature, the employer shall:
(a) provide the workers’ representatives concerned in good time with relevant information including the reasons for the terminations contemplated, the number and categories of workers likely to be affected and the period over which the terminations are intended to be carried out;
(b) give, in accordance with national law and practice, the workers’ representatives concerned, as early as possible, an opportunity for consultation on measures to be taken to avert or to minimise the terminations and measures to mitigate the adverse effects of any terminations on the workers concerned such as finding alternative employment’ (ILO, 1982).
Information Campaigns: At the moment, the uncertainty surrounding AI systems in the workplace is fuelling distrust in the technology (Chakravorti, 2024). Campaigns that inform workers not only of the capabilities of implemented AI systems and how to utilize their potential, but also about opportunities to upskill, are essential moving forward. Regardless of how these campaigns are run, giving workers more accessible information would go a long way towards realizing the ‘human centric’ ideals of the AI Act – ‘so that people can trust that the technology is used in a way that is safe and compliant with the law, including the respect of fundamental rights’ (European Commission (2021).
(*) Christo Pretorius graduated with a MSc in International Public Policy and Diplomacy from University College Cork and was the first student to receive a postgraduate “Student of the Year” award from the Department of Government. His dissertation was published and acquired by the Bar of Ireland’s Law Library and has gone on to support Irish policy makers. Stemming from his undergraduate in Ancient and Medieval History and Culture from Trinity College Dublin, his research interests include the mechanisms for authoritarian power and control, and democratic backsliding, particularly when viewed with a historical lens.
References
— (1982). C158 – Termination of Employment Convention. International Labour Organization. 1982 (No. 158). https://normlex.ilo.org/dyn/normlex/en/f?p=NORMLEXPUB:12100:0::NO::P12100_ILO_CODE:C158 (accessed on August 18, 2024).
— (1997). Organisation of Working Time Act 1997. Irish Statute Book. https://www.irishstatutebook.ie/eli/1997/act/20/section/23/enacted/en/html#sec23 (accessed on August 17, 2024).
— (2012). Charter of Fundamental Rights of the European Union. 326/02. http://data.europa.eu/eli/treaty/char_2016/oj(accessed on August 18, 2024).
— (2020) Special meeting of the European Council (1 and 2 October 2020) – Conclusions. General Secretariat of the Council to Delegations. Brussels. https://www.consilium.europa.eu/media/45910/021020-euco-final-conclusions.pdf
— (2021). Coordinated Plan on Artificial Intelligence 2021 Review. European Commission. https://digital-strategy.ec.europa.eu/en/library/coordinated-plan-artificial-intelligence-2021-review (accessed on August 19, 2024).
— (2021). Document 52021PC0206: Proposal for a Regulation of The European Parliament and of the Council Laying Down Harmonised Rules on Artificial Intelligence (Artificial Intelligence Act) and Amending Certain Union Legislative Acts (COM(2021) 206 final). European Commission. https://eur-lex.europa.eu/legal-content/EN/ALL/?uri=CELEX:52021PC0206 (accessed on August 17, 2024).
— (2022). Report on artificial intelligence in a digital age (2020/2266(INI)). European Parliament. https://www.europarl.europa.eu/doceo/document/A-9-2022-0088_EN.html (accessed on August 17, 2024).
— (2024). Artificial intelligence. European Commission. https://www.consilium.europa.eu/en/policies/artificial-intelligence/ (accessed on August 17, 2024).
— (2024). Corrigendum to Regulation (EU) 2024/… of The European Parliament and of the Council of laying down harmonised rules on artificial intelligence and amending Regulations (EC) No 300/2008, (EU) No 167/2013, (EU) No 168/2013, (EU) 2018/858, (EU) 2018/1139 and (EU) 2019/2144 and Directives 2014/90/EU, (EU) 2016/797 and (EU) 2020/1828 (Artificial Intelligence Act) (cor01). European Parliament. https://www.europarl.europa.eu/doceo/document/TA-9-2024-0138-FNL-COR01_EN.pdf
— (2024) The future of AI in the European Union: a chat with the AI Act’s co-rapporteur Brando Benifei. Publyon. https://publyon.com/the-future-of-ai-in-the-european-union-a-chat-with-the-ai-acts-co-rapporteur-brando-benifei/ (accessed on August 17, 2024).
Acemoglu D. & Restrepo P. (2021). Tasks, Automation, and the Rise in US Wage Inequality (NBER Working Paper 28920). National Bureau of Economic Research. https://doi.org/10.3386/w28920
Al Naam, Y.A.; Elsafi, S.; Al Jahdali, M.H.; Al Shaman, R.S.; Al-Qurouni, B.H. & Al Zahrani E.M. (2022). “The Impact of Total Automaton on the Clinical Laboratory Workforce: A Case Study.” Journal of Healthcare Leadership. 14, pp. 55-62. https://doi.org/10.2147/JHL.S362614
Chakravorti, B. (2024). AI’s Trust Problem. Harvard Business Review. https://hbr.org/2024/05/ais-trust-problem (accessed on August 17, 2024).
Gmyrek P., Berg J., & Bescond D. (2023). Generative AI and jobs: A global analysis of potential effects on job quantity and quality (ILO Working Paper 96). International Labour Organization. https://doi.org/10.54394/FHEM8239
Hawksworth J.; Berriman R. & Goel S. (2018). Will robots really steal our jobs? An international analysis of the potential long term impact of automation. PWC. https://www.pwc.co.uk/economic-services/assets/international-impact-of-automation-feb-2018.pdf
Jensen, C.L.; Thomsen, L.K.; Zeuthen, M.; Johnsen, S.; Jashi, R.E.; Nielsen, M.F.B.; Hemstra, L.E. & Smith, J. (2024) “Biomedical laboratory scientists and technicians in digital pathology – Is there a need for professional development?” Digital Health. 10. https://doi.org/10.1177/20552076241237392
Kochhar, R. (2023) Which U.S. Workers Are More Exposed to AI on Their Jobs?. Pew Research Center. https://www.pewresearch.org/social-trends/wp-content/uploads/sites/3/2023/07/st_2023.07.26_ai-and-jobs.pdf
Pretorius, P.C. (2023) “Can Irish Industrial Relations Still be Called ‘Voluntarist’? An investigation into Irish Employment Law”. Irish Employment Law Journal, 20(1), pp. 11-21.
von der Leyen, U. (2019). A Union that strives for more: My agenda for Europe: political guidelines for the next European Commission 2019-2024, Luxembourg: Publication Office of the European Union.https://commission.europa.eu/system/files/2020-04/political-guidelines-next-commission_en_0.pdf
Wold, J.F. (2024). “Von der Leyen gives nod to €100 billion ‘CERN for AI’ proposal.” Euroactiv. July 25, 2024.https://www.euractiv.com/section/digital/news/von-der-leyen-gives-nod-to-e100-billion-cern-for-ai-proposal/ (accessed on August 10, 2024).
