By Valerio De Stefano, Osgoode Hall Law School
In March 2024, The European Union’s institutions reached an agreement to adopt a European Directive on Platform Work. To understand the content of this Directive, it is essential to know what European directives are and how they are adopted. I’ll try to explain this without boring non-European readers to death.
An EU directive is a binding legal act of the EU. Directives are primarily addressed to Member States, which are given a certain period (usually two years) to implement their provisions into their national laws. Member States are held accountable if their national legislation does not faithfully comply with its content in due course. In some cases, private individuals can also invoke the provisions of a directive directly against Member States (but not against other private parties).
Adopting a legal instrument like an EU directive is a lengthy and complex process involving three main actors. The first is the European Commission, the executive body of the EU (composed mainly of unelected officials but led by individuals chosen by political bodies and thus indirectly representative). The second is the European Parliament, composed of members elected by all citizens of the European Union. The third is the Council, which comprises representatives of the governments of the individual Member States of the European Union (currently 27 countries). Therefore, an instrument like a directive must be deemed satisfactory by various bodies representing very different political interests. This is why its adoption can be lengthy and complex. This is also the point where you might start feeling a bit sleepy, but hang in there because this is where the important stuff begins.
The adoption of the Platform Word Directive perfectly reflected all the complexity I just described. Some political forces and national governments have proved to be particularly sensitive to the lobbying of platforms. Other governments, instead, pressed for common EU legislation and stronger protection for workers. Still, others are traditionally sceptical about adopting European labour regulations regardless of their content.
Furthermore, in drafting the various versions of the Directive, account had to be taken of the numerous rulings that various national courts have made in different Member States on the issue of the employment status of platform workers and their protection. Many national European courts have indeed ruled that workers from various platforms (e.g., Uber, Uber Eats, Glovo, etc.) are employees or should be protected as such. A minority of judicial decisions, however, have considered platform workers to be independent contractors.
In 2021, the European Commission proposed a draft Directive on Platform Work with two main components. The first was a rebuttable presumption of employment for platform workers. The draft identified five criteria indicating the presence of direction and control of workers by platforms. The presumption would have been triggered when two out of five of those criteria were present. This part of the Directive was the most contested.
For many Member States, the presumption proposed by the Commission was too stringent and would have risked causing an undue legal reclassification of workers even when they were genuinely independent. The Member States proposed a common draft that required at least three indicators of direction and control to trigger the presumption of subordination. However, even the Commission’s draft was insufficient to protect workers for the European Parliament. The Commission’s criteria were considered too generic and easy for platforms to circumvent. Elsewhere, I wrote that those criteria were inadequate and easy to circumvent and would have had the paradoxical effect of making it more difficult to reclassify workers as employees in various Member States.
The negotiation of the presumption provisions stalled the adoption of the Directive for several months, so much so that it was practically taken for granted that the instrument would not be approved. This would have left millions of platform workers within the European Union without common protections. Surprisingly, an agreement was reached at the eleventh hour. Even so, the agreement was reached with the opposition of France and the abstention of Germany, something highly unusual to any consequential EU decision (in case you are wondering, yes, the Directive will also bind these two countries).
According to the agreement, there will still be a rebuttable presumption of employment status. Unlike what the Commission and the European Parliament wanted, the presumption will not be defined by the EU but will be established by individual member states on the basis of common requirements. The first common element is that the presumption must be based on facts, not legal indicators. As I explained elsewhere and mentioned above, a presumption based on legal indicators risks complicating things for workers trying to be reclassified as employees.
The Directive also specifies that the presumption must facilitate the reclassification of platform workers as employees and cannot be used as an excuse to dilute the precedents of national courts that had already established that platform workers were employees. While this is obviously a compromise compared to the more ambitious proposals of the European Parliament, the final text is less risky than the initial drafts of the Commission and the Member States. National legislation will establish the contents of the presumption but will not have carte blanche, and the Court of Justice of the EU (CJEU) will be able to intervene when it deems that national laws are too lax. One thing you should know about the CJEU is that you don’t want to mess with it.
The second fundamental Chapter of the Directive concerns algorithmic management. This is undoubtedly the most innovative and ground-breaking part of the instrument.
Under the Directive, platform workers, even when genuinely independent, will receive adequate and understandable information about the algorithmic systems used to hire, direct, monitor, and discipline them. Systems monitoring worker performance must be transparent, and any system used to make decisions that have significant consequences on working conditions (such as wages, working hours, dismissals, etc.) must be carefully scrutinized. Platforms will be obliged to assess, together with the representatives of platform workers, the risks of discrimination that may arise from the use of algorithms. Platforms must also ensure that algorithms do not push workers to adopt an unsustainable work pace that puts them at physical or psychological risk.
Workers will also be able to request an explanation about – and request a human review of – relevant decisions made through algorithms that affect them and the rectification of those decisions that infringe on their rights. Moreover, and this is an essential protection that also applies to self-employed workers, any decision that has the effect of causing job loss, such as dismissal, suspension or closure of the account, cannot be made by algorithmic systems but must be made by human subjects.
Worker representatives will also have the right to be informed and consulted regularly about the nature of the algorithms and how they work. In this regard, it is essential to note that European laws do not require stringent requirements to be met for the certification of unions and other worker representatives before trade union rights are triggered. In many cases, the presence of worker representatives is automatic.
Two further plusses of the Directive must be added. Firstly, many protective provisions apply to all “persons performing platform work,” regardless of whether they are employees or self-employed. Secondly, equally important, the Directive concerns all forms of platform work. Unlike many national legislations, the Directive does not just apply to a limited sector of platform work (e.g., food-delivery riders, Uber drivers, etc.). This should facilitate the protection of those platform workers who have so far been neglected by attempts to regulate platform work (e.g. crowdworkers and other online workers, domestic workers, and clerical workers).
The Directive adds to a series of other measures that have already been adopted or will soon be adopted. For example, the European Commission has already adopted guidelinesthat allow many solo self-employed workers to access collective bargaining under current antitrust laws. Furthermore, EU institutions have also recently reached an agreement on the so-called “EU AI Act”. This is a general regulation of the EU on artificial intelligence. The regulation provides some essential protections and qualifies AI systems used to hire or manage workers as high-risk systems, triggering additional protection requirements. It also prohibits using AI tools to detect workers’ emotional states, requires worker representatives to be informed upon the adoption of AI tools at work, and authorizes Member States to maintain or introduce stricter rules on this point at the national level.
Although particularly innovative, even the Chapter on algorithmic management of the Platform Work Directive and the labour provisions of the EU AI Act could have been strengthened. In particular, unlike what happens at the national level for other electronic monitoring tools, these European instruments do not introduce an obligation of collective bargaining or veto powers by worker representatives before introducing electronic monitoring tools. There is a problem with EU competencies in adopting these more stringent obligations. However, it is paradoxical that in many EU countries, introducing surveillance cameras in workplaces will still require a collective agreement when these new European legislations do not mandate this requirement when employers introduce monitoring and management tools that can be much more invasive and riskier. EU Member States will, therefore, probably be called upon to adopt even more effective protective standards.
There is no doubt, however, that compared to what happens in other industrialized countries, the instruments discussed so far are much more innovative and protective. The protections introduced or proposed in Canada at the moment pale in comparison. In Ontario, the Employment Standards Act was amended months ago to mandate adopting a unilateral policy and simple individual information obligations when employers with more than 25 employees introduce electronic monitoring tools. These are merely procedural obligations that do not provide any substantive rights for workers or impose any substantive obligations for employers. The same is happening at the federal level regarding Bill C27 on artificial intelligence. We can hope these new European legislations constitute a wake-up call for the various Canadian regulators.
More realistically, however, it is time for Canadian unions and labour activists to take much more interest in the risks that algorithmic management poses to all workers, unionized and non-unionized, and actively campaign for greater protections.