Skip to content

Two steps forward, one step back: the EU’s plans for improving gig working conditions

A critique of the EU’s proposed approach to employment status and algorithmic management for platform workers

Antonio Aloisi , Despoina Georgiou

7 April 2022

Reading time: 14 minutes

Food riders protest against new lockdown restriction enforcement and new contract

Aware of the precarious situation of platform workers, in December 2021, the European Commission published its proposal for a new Directive on improving working conditions in the gig economy, accompanied by guidelines on collective bargaining for some categories of self-employed workers. The proposed Directive follows growing concern about issues related to the status, protection and rights of those working in the platform economy, as well as an increasing number of legal cases and worker protests.

This blog post analyses critically the merits and shortcomings of the proposed legislative instrument, at a time when the discussion among EU co-legislators is still unfolding.

As in other domains of the digital economy, this proposal positions Brussels institutions as ‘leaders in the regulatory arena’ that might set a trend. It is therefore necessary to understand the logic of the proposed legislation in the context of existing case law in countries like Italy, Spain, France and the UK. Moreover, among other things, the proposed text is particularly innovative as it concerns the practice of management by algorithms, which is increasingly common in all forms of work. In recent years, platform work has grown explosively in Europe and elsewhere. In the EU, it is estimated that over 28 million people have their work mediated by digital platforms. Digital platform work has been described as boosting productivity, enhancing flexibility and facilitating labour market integration by providing a source of income for marginalised groups. However, recent data questions the viability of the model, which has widespread negative externalities and has sparked outrage for the low earnings and poor working conditions.

Those with recognised status as ‘worker’ in the EU benefit from protective legislation regarding equal pay, working time (holidays, rest periods, overtime pay), maternity and paternity leave and pay, health and safety, information, consultation and collective redundancies. Empirical research shows that up to 5.5 million people working through digital labour platforms in the EU today are at risk of employment status misclassification. This means that delivery couriers, cleaning staff and online freelancers are under-protected and have limited access to the vast bulk of employment and social security entitlements. The proposed Directive, acclaimed as a bold move, aims to ensure the correct determination of workers’ employment status, improve transparency and promote accountability in algorithmic management.

In this post, we address the limitations of the mechanisms in the Directive for triggering the presumption of employment status, which may constrain its scope of application. We claim that the set of indicators needing to be met may upset domestic legal advances in national European states, when it comes to the notion of ‘control’, to which employment status is indexed. We also find that, albeit forward-looking, the proposals on algorithmic management leave room for unscrupulous business practices and fall short of providing mechanisms of co-determination.

Much is known about the underlying contradictions of platform work. Nine out of ten platforms active in the EU today classify people working through them as self-employed. And yet, many of these persons are subordinate to traditional forms of power, which is exerted in subtle and deceptive ways. Platforms often set minimum quality standards for service, strict branding requirements and terms for payment. They also use automated systems to assign tasks, organise shifts, monitor workers and determine their access to jobs through internal rankings. In short, platform work has been the harbinger of ‘algorithmic management’, a set of practices whereby decision-making functions are assigned to AI-driven software and apps.

In addition, platforms have also found ways to pass down financial and commercial risks to workers. Under many contracts, workers are called upon to shoulder various risks related to payment, investment and maintenance, health and safety, redeployment, and insurance. Undoubtedly, if these technological and organisational developments do not take place in a regulated and responsible manner, they could lead to the erosion of workers’ rights.

Status update

Who are the beneficiaries of the new EU legal tool? Promisingly, workers who operate both in a certain location and online are included. Given the hurdles in terms of enforcement and the scarce visibility of (online) ‘crowd-workers’, this community has been often neglected in other policy measures. However, the proposed Directive has a strikingly broad scope, in terms of who is included and, unlike some laws introduced in domestic jurisdictions that concern only a particular sector, typically food- or grocery-delivery services in urban areas (e.g., the Spanish ‘Ley Riders’), the proposed Directive covers platform workers engaged in all types of work.

According to Article 1(2), the minimum rights laid down in the Directive apply to ‘every person performing platform work in the EU who has, or who, based on an assessment of facts, may be deemed to have, an employment contract or relationship as defined by the law, collective agreements or practice in force in the Member States with consideration to the case-law of the Court of Justice’ (CJEU).

Consistent with its purpose, the Directive reinforces the ‘primacy of facts’ principle, according to which the contractual designation can be defeated by an appraisal of factual circumstances. This is a significant step forward, even compared to the 2019 Directive on transparent and predictable working conditions, which failed to design a more inclusive scope and adopted a ‘hybrid’ formula, in which varied domestic definitions of ‘workers who has an employment contract or employment relationship’ have to be read in accordance with CJEU case law.

To help tribunals make their factual assessment, Article 4 of the proposed Directive introduces a legal presumption of ‘worker’ status for persons who are controlled by a platform. This ‘control’ is considered to be in place when two out of the five following criteria apply:

  • the platform determines, or sets upper limits for the level of remuneration;
  • the platform imposes specific binding rules with regard to appearance, conduct or performance;
  • the platform supervises performance or verifies the quality of work, including by electronic means;
  • the platform restricts the freedom, including through sanctions, to organise one’s work, in particular the ability to choose working time or periods of absence, to accept or to refuse tasks or to use subcontractors or substitutes;
  • the platform restricts the possibility to build a client base or to work for any third party.

Platforms can challenge the presumed worker status by proving that the contractual relationship in question is not an employment relationship. In an effort to secure effective implementation, the Directive also imposes transparency duties on Member States. They are required to make this ‘blueprint’ available to workers in a clear, comprehensible and easily accessible way. Simultaneously, given the inherent information asymmetry, workers must be supported in understanding and defending the presumption of employment status, while administrative controls and inspection must be strengthened. Nonetheless, the presumption does not itself provide a universal ‘fix’ to the various challenges posed by the ‘gigification’ of the labour market. In fact, there is significant room for improvement from a personal, procedural and substantive standpoint.

Litigation legacies

While the list of proxies for control includes several factors that have been already used by courts to reclassify platform workers, the requirement that two out of five criteria need to be established to trigger the presumption could prove too restrictive. The presence of any one criterion should be sufficient, otherwise there is a risk of a de facto regression around the corner.

On a closer inspection, thanks to extensive litigation around the employment classification of platform workers, case law in countries such as Italy, Spain, France and the UK is already used to an evolutionary and dynamic notion of control, which is less stringent than the one demanded by the Directive. A narrow EU-wide definition could unintentionally set a higher threshold than domestic legal criteria. It will be crucial to make sure that the bar for the presumption of employment status is lowered in the final version of the text.

Some of the indicators of control (substitution or non-exclusivity clauses) could have a boomerang effect, reinforcing the view that the platform worker’s ‘flexibility’ to accept or refuse tasks, or to use substitutes, denotes a level of autonomy, even though the person has no real choice over these matters. As demonstrated in the Yodel issued by the CJEU in April 2020, these criteria could impair the application of the ‘primacy of facts’ principle, embracing a formalistic reading of contracts, sometimes justified by the peculiar nature of this business model. These shortcuts have been already used by Deliveroo and other companies in the UK to prevent reclassification. EU institutions could seize this opportunity to get rid of ‘terms and conditions ﷟doublespeak’ and avoid legal arbitrage.

Moreover, while the Directive does not provide the criteria for rebuttal, it states that a person can be self-employed ‘even though the digital labour platform controls the performance of work on a given aspect’ (Recital 28). This means that a person might be found to be ‘self-employed’ even though they are under some level of control from the platform. This is because ‘control’ is not the only criterion used by the CJEU and the domestic courts to classify individuals. Other criteria such as ‘the assumption of business risks’,1 ‘the exclusivity in the relationship between the parties’, and the ‘working persons’ level of integration into their putative employer’s business unit’ have often been taken into account. If these criteria are used, many who provide their services through digital labour platforms could be deemed to be ‘self-employed’, even though they are under the control of platforms. This could make it easier for platforms to rebut the presumption than it first appears, while opening the door to greater legal uncertainty, resulting in further litigation.

To these pitfalls and problems must be added the carve-out for start-ups, to which the new rules do not apply, and a provision preventing retroactive effects before the Directive’s transposition date. Such measures unjustifiably remove protections from vulnerable platform workers and should not be included.

The Commission estimates that only 1.7 to 4.1 million persons will be reclassified as ‘workers’ as the result of the Directive. This means that many in need of protection will not be covered by the proposed instrument, whose primary goal is precisely to rectify employment status. In light of recent legal developments, EU countries should take an anticipatory stance, avoid further labour market segmentation and update existing instruments to achieve the goals of the Directive sooner than envisaged.

Algorithmic bosses

The most modern chapter of the draft involves the regulation of algorithmic management. A broad right to access information is guaranteed for platform workers as regards:

  • digital monitoring systems that supervise or evaluate performance;
  • automated decision-making systems (ADMS) that are implemented to take or support decisions significantly affecting working conditions (for example, access to work assignment, earnings, working time, promotion or account termination).

When they implement surveillance systems, platforms must disclose these and the types of activity monitored, including customer activity. Likewise, besides informing workers of their existence, platforms must present the types of decisions taken or supported by automated decision-making systems (ADMS), the parameters factored in and their relative weight, and the grounds for decisions to restrict, suspend or terminate a worker’s account (Article 6).

These provisions are in an ‘ambivalent’ relationship with the EU General Data Protection Regulation (GDPR).2 On the one hand, they complement and advance the existing standard by including decisions that are merely supported (as opposed to fully carried out) by ADMS. They also solve the long-standing quandary on the existence of a right of explanation in the most progressive way, so broadening the interpretation of the GDPR. On the other hand, the Directive condones the adoption of such pervasive organisational tools and defines a set of prohibited practices that is overly narrow (those based on the emotional and psychological state of the worker, health information, private conversations with worker representatives and data not related to work).

Management by algorithms is among the reasons why platform work is here to stay, irrespective of its current dimensions. The wildest experimentations conducted in this context are becoming widespread practices that have been further boosted during the two years of the pandemic. Work-from-home arrangements and the need to secure compliance with exceptional health and safety guidelines have pushed the normalisation of ‘panopticon’ devices and software to the extreme. Regrettably, ‘soft’ data protection rights are recognised by the proposed Directive exclusively to platform workers. A more universalistic intervention would have been welcome as a step towards curbing the augmented power of digital bosses.

Notably, digital labour platforms shall make the information mentioned above available to platform workers’ representatives and national labour authorities upon their request (Article 6). It is positive that the proposal envisages collective channels for information and consultation, increased transparency and redress (Articles 9 to 15). However, workers’ representatives and trade unions must have a seat at the table where decisions are made. They should be empowered to co-determine the adoption of technological tools with monitoring and decision-making potential, as is already the case in several national models, by means of comprehensive, adaptive and anticipatory collective agreements.

Moreover, the Directive stipulates that platform workers who are genuinely self-employed should enjoy transparency rights as regards algorithmic monitoring and management. While initial reactions have hailed this as a promising development at a moment when workplaces are ever more datafied, it is now becoming clear that the provision on extending the protections from algorithmic harms to the self-employed is not immune from risk.

All over Europe, according to the decisions of national courts, such a degree of monitoring and control would be sufficient for employment status. However, the Directive accepts that those who are self-employed could be monitored and managed by automated systems. Such an ambiguity could be open to all sorts of opportunistic strategies, often coupled with the tactic of using intermediaries in order to dilute responsibilities and discourage grievances.

In short, the text risks being self-defeating: how can it be acceptable that platform workers are subject to intense algorithmic management and yet still classified as self-employed, when control is the key trigger of the presumption of employment status? This contradiction deserves clarification to prevent unintended consequences.

The current formulation of data rights for self-employed platform workers betrays that the Directive is still embedding a misunderstanding of contemporary conditions of work. All workers (employed by a platform or not) are now likely to be subject to some form of algorithmic management and so would benefit from strengthened informational self-determination rights. Moreover, the fact that the Directive admits the possibility of a self-employed person being under such intense algorithm-driven management frustrates the intended purpose of the new legislative instruments.

The task ahead

EU institutions should not abandon their role in the regulation of technology-driven work arrangements. Contrasted with the California Assembly’s stronger presumptive system (although the fate of the AB5 model is still uncertain, after the referendum that quashed it), the standard-setting aspiration of the proposed EU Directive could be enhanced, particularly in light of the fluid activism prompted by the launch of the European Pillar of Social Rights.

By adopting a pragmatic approach, we hope this blog post has highlighted some contentious points that deserve answers from the EU co-legislators in the final text. Together with the Guidelines on collective agreements on the working conditions of solo self-employed persons,3 the Platform Work Directive upholds a sustainable digital transformation of work, à la Européenne, whereby genuine innovation and social rights are strengthened in harmony, which can be embraced and advanced at the level.



  1. C-413/13, FNV Kunsten Informatie en Media v Staat der Nederlanden [2014]; C-3/87, The Queen v Ministry of Agriculture, Fisheries and Food, ex parte Agegate Ltd [1989].
  2. Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95/46/EC (General Data Protection Regulation).
  3. European Commission, Guidelines on the Application of EU competition law to collective agreements regarding the working conditions of solo self-employed persons COM(2021) 8838 final
The future of regulation
AI policy