Platform Workers' Employment Status — When the App Gives the Orders

The List Posted in the Group Chat

In October 2018, a service launched where calling via an app summoned an 11-seater van complete with a driver. Legally, it was a "car rental service with driver arrangement" where a car rental company rented vans to members who joined the app while arranging a driver, and the driver recruitment posting stated it was for driving in the Seoul area (Seoul High Court 2022Nu56601). On July 12 of the following year, a notice was posted in the KakaoTalk group chat of the drivers operating these cars. It stated that starting July 15, work shifts would be reorganized and the number of vehicles adjusted to reduce personnel. Attached to the notice was a list of 22 drivers who would be dispatched moving forward, and it stated that those not on the list could not continue together (Seoul High Court 2022Nu56601). The entity that posted the notice was a staffing agency that recruited and supplied drivers. One person omitted from that list was a driver who had signed a "Driver Freelance Contract" with this agency that May and driven cars every weekend.

Three months later, this driver applied for relief, claiming that the notification was unfair dismissal. A question arises here. Can someone who signed a freelance contract also be dismissed? Because dismissal is something that happens to a worker, this question immediately transforms into whether this person is a worker. The dispute ended five years later at the Supreme Court (Supreme Court 2024Du32973). The judgment masked the company names as ○○ and △△, and this case is known as the driver case for the van hailing service "Tada."

This article centers on this Korean judgment and compares it with the answers provided to the same question by the United Kingdom, the Netherlands, France, New Zealand, South Africa, the European Union, and California, USA. The Korean judgment is translated from the original judgment text, and the judgments and legislation of other countries are translated only as far as stated by the paper abstracts describing them. The legal literature cited varies in rank. Preprints or studies covering only a single country are noted as such.

The Paperwork All Pointed to a Sole Proprietor

The common belief is simple. A person who gets work through an app is a free independent business owner, and the contract says so as well. A sole proprietor is someone who is not employed by others and takes on work at their own calculation and risk; in Anglo-American law, they are called an independent contractor. The Labor Standards Act is a law that sets the minimum baseline for working conditions such as wages, working hours, and dismissal, but this law does not apply to sole proprietors.

This driver also looked close to a sole proprietor on paper. The job posting stated that he could choose his own working days and that the pay was 10,000 won per hour. In reality, without a base salary or fixed pay, he received 10,000 won per hour of driving time and special commissions based on his grade. Business income tax, rather than earned income tax, was withheld from his compensation, and he was treated as a sole proprietor under social security systems like employment insurance (서울고법 2022누56601). Because he could choose his working days, time slots, and garage to apply for dispatch, he did other work during the week and drove only on Saturdays and Sundays (대법원 2024두32973).

He had even said so himself. Six days after the notice of headcount reduction, on July 18, he wrote in a group chat comparing the pros and cons of freelance and employment contracts, stating, "We are proud freelancers who gave up employee benefits and were hired because they needed us" (서울고법 2022누56601).

The first ruling was also on the side of common belief. The Labor Relations Commission is an administrative agency established to provide faster and simpler relief than lawsuits for labor disputes such as unfair dismissal (대법원 2024두32973). In December 2019, the Seoul Regional Labor Relations Commission rejected the driver's request for relief on the grounds that he was not an employee under the Labor Standards Act (dismissal). The National Labor Relations Commission, which handled the re-examination, overturned that decision in May 2020. It viewed the driver as an employee, the employer that substantially directed and supervised him as the company that rents out the vehicles, and the headcount reduction carried out without written notice as an unfair dismissal. The application against the company operating the app and the manpower supply agency was dismissed (서울고법 2022누56601).

The company that rents out the vehicles filed an administrative lawsuit asking to cancel this re-examination decision. Thus, the court positions were set as follows. The plaintiff is the vehicle rental company, and the defendant is the Chairperson of the National Labor Relations Commission. The driver entered as an auxiliary participant supporting the defendant's side. Departing from the conclusion of the first instance at the Seoul Administrative Court (July 2022), the second instance at the Seoul High Court in December 2023 canceled the first-instance ruling and did not accept the company's claim (서울고법 2022누56601). The company asked the Supreme Court for a re-evaluation, and the Supreme Court dismissed the appeal on July 25, 2024. It was the unanimous opinion of four Supreme Court Justices (대법원 2024두32973).

The company side also pointed out that an employee of the manpower supply agency sent a text message notifying weekend driving opportunities five days after the notice, on July 17, and that the driver refused, but the lower court stated that this text message could not be seen as a withdrawal of dismissal or a request for reinstatement (서울고법 2022누56601).

Contracts, taxes, insurance, and even his own words—the paperwork pointed to one side. What the court asked was: what actually happened outside of those papers?

What the Supreme Court Sees

Worker status is the quality of whether a person falls under the workers protected by the Labor Standards Act. The sentence the Supreme Court took as its starting point is an established legal doctrine. Whether someone is a worker under the Labor Standards Act is judged not by the formal nature of the contract—whether it is an employment contract, a service contract, or a mandate contract—but by whether they actually provided labor in a subordinate relationship for the purpose of earning wages (Supreme Court 2024Du32973). This means the actual relationship comes before the name attached to the contract.

A subordinate relationship is one where the worker performs work as determined by the other party, receives their direction and supervision, and lacks the independence to run it as their own business. The ruling lists the indicators used to gauge this. It looks at whether the employer determines the contents of the work and whether the worker is subject to service regulations, and whether they receive substantial direction and supervision in the work process. It also looks at whether the employer determines the working hours and location and whether the worker is tied to them. It examines whether they can run a business on their own account—such as by preparing work tools themselves or hiring others to work in their place—and whether they bear the risks of profit and loss. Factors such as whether the remuneration is compensation for the work itself, whether the relationship is continuous, and whether the worker is exclusively bound to a single employer are also included. The ruling states that these various economic and social conditions should be assessed comprehensively (Supreme Court 2024Du32973).

A single following proviso takes aim at the documents from the previous section. Circumstances such as whether a basic or fixed salary was set, whether earned income tax was withheld at source, or whether a person is recognized as a worker under social security systems are largely prone to being arbitrarily determined by the employer using their economically superior position. Therefore, the ruling notes that worker status should not be easily denied merely because these points are not acknowledged (Supreme Court 2024Du32973). Hourly pay, withholding business income tax instead of earned income tax, and treatment as an independent contractor are all things that a company can write on a document. The lower court's ruling in this case—specifically, the judgment by the Seoul High Court, which tried this case immediately prior—addressed the individual's statements as follows. The status of a worker is determined by the reality of the relationship, and is not decided by whether the worker themselves recognized that they were a worker (Seoul High Court 2022Nu56601).

The Supreme Court added a sentence for platforms here. The ruling defines an online platform as an electronic information processing system that allows two or more users to exchange things related to work. Then, when judging whether platform workers are workers, it instructs that the above indicators should be appropriately applied by taking two characteristics into consideration. One is a business structure where the person doing the work and the person using the work are connected through a platform, reducing the need to directly sign individual labor contracts. The other is labor management where the platform's algorithm or various business participants are involved in determining how work is distributed and performed (Supreme Court 2024Du32973). The Supreme Court stated that the same criteria for subordination should be used for platform labor, but applied in consideration of algorithms and the structure where multiple companies share and handle work.

What this sentence does not determine is also clear. After clarifying the criteria, the ruling applied them to the facts of a single service and a single driver to reach a conclusion. This is a ruling by a panel consisting of four Supreme Court Justices, and it is not a declaration that anyone who works via a platform is a worker. For other apps and other jobs, the same indicators must be examined one by one all over again.

A Driver's Saturday

We follow the facts painted by the judgment through one day in the life of this driver. The company operating the app sent a dispatch table listing depots and operating hours to partner agencies every week. The staffing agency collected preferred work days, depots, and start and end times for the following week from its affiliated drivers via a Google Form by midnight every Thursday (Seoul High Court 2022Nu56601). After pairing these applications, the app operator sent the vehicle's smart key to the dispatched driver through the app. Dispatching was completed in this manner (Supreme Court 2024Du32973).

When Saturday arrived, he arrived at the depot at the designated start time, got into the dispatched car, and pressed the "Start Work" button on the app. After that, he went to a waiting area designated by the app and waited for calls. When a call came in, he had to decide whether to accept it without receiving information to gauge the content of the work, such as the destination, and passengers could not choose their drivers. Going to work, whether to accept calls, travel routes, breaks, and finishing work were all automatically recorded and managed by the app (Supreme Court 2024Du32973). According to the records confirmed by the lower court, he drove 15 times without missing a single time from noon to 10 PM on the Saturdays and Sundays he was scheduled to work between May 26 and July 14 (Seoul High Court 2022Nu56601).

He too had freedom. The Supreme Court noted that he could choose whether to accept calls and take breaks, and could end work at the time he wanted. However, other facts follow in the same paragraph. The staffing agency announced that it would impose educational disciplinary measures if he did not take calls, calling it 「instructions sent down from △△」. The educational materials contained phrases such as 「Penalties in case of refusal or non-acceptance of dispatch (calls)」 and 「Unaccepted cases will be reflected in the total monthly unaccepted cases, which may cause disadvantages in personnel evaluations」, and an initial app screen popped up with a notice that non-acceptance would be applied unfavorably to personnel evaluations. The app operator implemented a driver tier system starting in February 2019, giving different special commissions according to grades, and the evaluation items included the number of trips, driving distance, number of days worked, number of unaccepted and canceled calls, and number of departures from waiting areas (Supreme Court 2024Du32973). The lower court noted that the special commission rates by grade were 10%, 8%, 5%, and 0% (Seoul High Court 2022Nu56601).

The Supreme Court tied these facts together and made a judgment. It stated that as long as driving details were recorded in the app and sanctions or the loss of special commissions were scheduled accordingly, it was difficult to see that he could freely choose whether to accept calls, take breaks, and finish work. Regarding receiving work assigned by the app without knowing the destination, it judged that since the online platform selected the person to work, distributed work items, and designated and controlled how to work, it could not be seen that full freedom of choice was given (Supreme Court 2024Du32973). As a number supporting this judgment, the lower court pointed out that the ratio of non-acceptance among the total dispatches of freelance drivers from May to July 2019 remained at 0.83 to 1.29% per month (Seoul High Court 2022Nu56601).

The Supreme Court also acknowledged as is the point that he could choose the days of the week. However, since he could only drive by accepting a dispatch application and having a car handed over, it viewed that the working hours and workplace were ultimately determined by the app operator acting on behalf of the car rental company (Supreme Court 2024Du32973). What the driver could choose went only as far as applying for the days he wanted to work, and the app operator decided whether to accept that application and hand over the car.

Signs that should be present if he were a business owner were not visible. He could not have someone else do the driving for him, nor could he earn more money by picking up passengers other than the passenger designated by the app. The car and supplies all belonged to the car rental company, and costs such as car wash fees and fuel costs were also paid by that company. He did not bear the risk of making a loss. Since remuneration was received in proportion to working hours regardless of the quality of the work, it was said that it could be seen as compensation for labor itself. It stated that the facts that there was no base salary and earned income tax was not withheld were due to the characteristics of labor provision mediated by the platform, so it was difficult to place great significance on them. Regarding the point that he did other work during the week, since he could only drive for this service within the dispatched driving hours, it was difficult to view that exclusivity was low just because working hours were short (Supreme Court 2024Du32973).

Who Is the Employer?

Even if the answer comes out as a worker, one question remains: whose worker is he? There were three companies in this service. The car rental company owned the cars, paid 10% of the usage amount as a commission to the app operator, paid hourly driving service fees to the partner company, and kept the remaining income. This company also determined the freelance drivers' wages and work details. The app operator was acquired by the car rental company as a 100% subsidiary, created and ran the app, and made training materials and dispatch refusal guidelines to distribute to partner companies. It created and sent attendance management reports using the drivers' driving history, and sometimes requested that measures such as training be taken and the results reported. The manpower supply company that signed a contract with the drivers was a business licensed for worker dispatching, which recruited and supplied drivers, received driving service fees, delivered remuneration to the drivers, and took the price difference as a commission (Supreme Court 2024Du32973).

The Supreme Court stated that when finding an employer, one should base it on the actual employment relationship regardless of the form of the contract or the contents of related laws and regulations. Based on that standard, the app operator was evaluated as having acted on behalf of the car rental company rather than performing a part of the service independently. The manpower supply company used the training materials received from the app operator almost as they were, and was seen as lacking data or means to separately determine the drivers' specific work details or independently manage and supervise their work performance. The conclusion is this: the drivers are workers under the Labor Standards Act who provided labor to the car rental company in a subordinate relationship, and their employer is the car rental company (Supreme Court 2024Du32973). The company that wrote the contract and the company that made them work were different, and the court found the one that made them work to be the employer.

This structure left its traces on other issues in the lawsuit as well. An application for relief from unfair dismissal must be made within 3 months from the date of dismissal. On October 7, 2019, the driver first applied against the app operator, and through a process of adding and removing the manpower supply company, only added the car rental company as the opposing party on December 3 (Seoul High Court 2022Nu56601). The company argued that the deadline had passed. The Supreme Court did not accept this argument, stating that it should be viewed based on the timing of the initial application, and one of the reasons for that is as follows. As employment types diversify, it is often difficult for workers to accurately identify who their employer is from the beginning, and in such cases, they are more likely to be in a economically vulnerable position (Supreme Court 2024Du32973). The court recognized in its procedural rules that even finding one's own employer is often difficult for the working person.

Even among drivers driving the same car, contracts were split. According to the lower court ruling, as of March 2020, out of 11,444 drivers for this service, 1,368 were dispatched drivers. The lower court judged that there was no meaningful difference in work content and command-and-supervision between the dispatched drivers recognized as workers by the company and the freelance drivers (Seoul High Court 2022Nu56601).

The scope stated by the ruling is up to here. It determined worker status and the employer from the structure of this service and the factual relations of this driver. This ruling determined nothing regarding food delivery workers or workers of other hailing apps.

When an Algorithm Is the Manager

In the Tada case, a substantial portion of instructions and supervision was delivered through app screens and driving logs. Researchers call this approach by a separate name. Algorithmic management is an approach where software rules take over the work of managers—such as task allocation, evaluation, compensation, and sanctions—in place of human beings. A study that qualitatively investigated the perceptions of Uber drivers and supplemented them with interviews of Uber executives and engineers summarizes that algorithms are used on two axes. One is matching, which pairs work and people, and the other is control, which monitors and tightens up work. On both axes, drivers experienced tensions surrounding task execution, compensation, and a sense of belonging, and responded to these tensions either like market participants or like organizational members (DOI: 10.25300/misq/2021/15333).

A qualitative case study examining food delivery labor in Australia views control as being multilayered rather than just algorithmic management. Three prominent features stand out: a technology base that peers into everything, using information asymmetry to narrow a worker's choices, and a dimly obscured performance management system. Workers' responses were limited and mostly remained at the individual level (DOI: 10.1177/0950017019836911). This looks just like the Korean drivers who had to decide whether to accept a call without knowing the destination. Because the country and the service differ between studies, we can only say this much: that they look alike.

Not all apps tighten the screws to the same degree. A study analyzing in-depth interviews with 55 food delivery platform workers and surveys of 955 found that while all platforms allocated and evaluated work through algorithmic management, the variation among platforms was large. The grocery delivery platform Instacart regulated workers' time and activity more strictly than other delivery platforms, leading the research team to call this "algorithmic dictatorship" (DOI: 10.1177/2378023119870041). When the intensity of control differs this much under the same name of "platform labor," the conclusion of one case cannot be transferred as-is to another app. A Turkish paper commenting on the UK Supreme Court ruling regarding the Uber case reaches the same conclusion in light of its own country's laws. It argues that the conclusion that everyone working through a platform is an employee is not possible, and that subordination must be assessed on a case-by-case basis according to criteria (DOI: 10.34246/ahbvuhfd.933999).

Whether algorithms eliminate control or hide it is a matter of debate. A comparative law paper from 2026 views platforms as performing the user functions of task allocation, compensation determination, performance monitoring, and connection blocking through algorithms that workers can neither see nor contest, instead of a visible supervisor. The author argues that algorithmic management does not eliminate user control, but rather moves it out of sight and hides it (DOI: 10.63108/vab.ldd.2.15). The sentence used by the Korean Supreme Court, "the platform selects the person to work, allocates tasks, and designates and controls the method of work," points in the same direction as this argument. However, the Supreme Court did not declare this as a general theory, but rather applied it to the facts of this specific case.

The Same Question, Different Answers in Different Countries

In a 2017 case reported from the UK, a first-tier employment tribunal ruled that Uber drivers are not independent contractors. However, the place where the drivers were put was not as employees. A worker is an intermediate status that UK law places between an employee and a self-employed person. UK employment law divides working people into three categories: employees, workers, and the self-employed (DOI: 10.61315/lselr.1047). By being judged as workers, the drivers received basic protections such as the minimum wage, paid leave, and protection from detriment when reporting whistleblowing, and at the time there were roughly 40,000 Uber drivers in the UK (DOI: 10.5553/eelc/187791072017002001011). In 2021, the UK Supreme Court unanimously confirmed that the drivers are not independent contractors but workers, and considered that a driver is working from the moment they log into the app and are ready to work (DOI: 10.5553/eelc/187791072021006002013).

Therefore, the UK Uber ruling is a judgment that placed drivers in an intermediate status and granted them basic protections, and it is different from a ruling recognizing them as employees. Even within the UK, food delivery platform Deliveroo's riders were not granted worker status by the High Court (DOI: 10.1177/1023263x211051833). Criticisms remained even after Uber drivers were recognized as workers. A paper in the student-edited academic journal LSE Law Review argues that the UK Supreme Court rulings on Uber and Deliveroo treat contractual freedoms—such as the right to send a substitute to work or logging into multiple apps at once—as decisive proof of autonomy. The criticism is that because of this, many platform workers who have such freedoms in their contracts are left in a "legal statelessness" that belongs to none of the three statuses (DOI: 10.61315/lselr.1047).

Continental Europe's conclusions are different again. Dutch courts ruled in 2021 that both Deliveroo riders and Uber drivers are employees, and the Amsterdam District Court considered that a "modern relations of subordination" exists between Uber and the drivers (DOI: 10.5553/eelc/187791072022007002009). France's Court of Cassation judged that an employer-employee relationship exists regarding delivery riders (DOI: 10.1177/1023263x211051833). The Court of Justice of the European Union, the European Union's highest court, stated in the 2020 case of UK courier company Yodel's delivery drivers that while employment status is for each country's courts to determine, these drivers appeared correctly classified as self-employed. The reason was that there was room for the drivers to determine task acceptance, competitor work, substitute labor, and work schedules, and the decisive factors were independence and subordination (DOI: 10.5553/eelc/187791072020005003018).

Other continents are divided too. The New Zealand Supreme Court judged in 2025 that Uber drivers are employees under the Employment Relations Act. The judgment still focused on the parties' intentions while being supported by the traditional criteria of control, organizational integration, and the reality of the business (DOI: 10.59403/126ekg4). In South Africa, the Labor Court in 2018 stated that a dispute resolution body's judgment viewing South African Uber drivers as employees of Uber's South African entity (Uber SA) was incorrect (DOI: 10.47348/ilj/v45/i1a2). This ruling dealt with who the employer is, so carrying it over to a ruling that established drivers are not employees would be mistaken.

Regarding the same Uber drivers, the UK saw them as an intermediate status, the Netherlands and New Zealand as employees, and the South African Labor Court did not accept the judgment viewing the drivers as employees of the local entity. The Court of Justice of the European Union in the Yodel case sided with self-employment. Within Europe, a trend of the lists of facts being examined converging is also visible. A preprint study analyzing case law from European Union member states views this as a trend of gradual harmonization, noting that courts increasingly look at functional markers of organizational integration rather than formal contract clauses or nominal autonomy (DOI: 10.2139/ssrn.5706383, preprint). This is a list that overlaps with dispatch, level systems, and fare collection examined by Korean rulings. The facts that courts look at are similar, but how much weight is placed on those facts varies by country and case.

When Legislation Answers — California and the European Union

Some places try to set answers through legislation instead of case-by-case court rulings. In California, USA, the California state law AB5 was created. The author, who studied the law and the field in California for 10 years, writes that AB5 was a law designed to make platform companies treat drivers as workers and provide safety nets like a living wage, unemployment insurance, workers' compensation, and health insurance. The author argues that platform workers who organized themselves were decisive in the passage of this law. And they write that large platform companies sponsored and succeeded in a ballot measure targeting AB5 in the middle of the COVID-19 pandemic, criticizing the implications of its passage for democracy (DOI: 10.1177/20319525211063111).

That ballot measure is Proposition 22 (Prop 22) on November 3, 2020. A ballot measure written by major tech companies like Uber and Lyft passed with 58% of the vote (DOI: 10.4324/9781003161875-23). A ballot measure aimed at forcing app-based drivers to be treated as workers won. The specific provisions of the ballot measure are not covered by the cited abstracts, so they are not transcribed here. Written from the perspective of worker organizing, the author of this piece views that after Prop 22, similar attempts were foreshadowed in other states and the federal legislature, and that threat frightened some unions into labor-management agreements that cement a third legal category (DOI: 10.4324/9781003161875-23). Questions remained even after Prop 22. In a California study combining 41 documents, 8 expert interviews, and 7 cases from October 2022 to May 2024, the concerns of drivers highlighted in expert interviews were fair pay, benefits, worker classification, and platform transparency regarding discipline such as account deactivation. The researchers concluded that minimum pay and account deactivation appeal procedures could improve working conditions (DOI: 10.1016/j.tbs.2025.101218).

The European Union chose a different path. In 2021, about 28.3 million people engaged in platform work in the European Union, and 5.5 million of them were subjected to algorithmic control to varying degrees. The European Commission estimated that a significant portion of platform workers classified as self-employed on paper were actually misclassified (DOI: 10.2139/ssrn.5706383, preprint). Misclassification is the practice of classifying people who are actually workers as independent contractors on paper. The core mechanism of the Platform Work Directive 2024/2831, adopted in October 2024, targets this misclassification. The employment presumption is a mechanism where, if certain signs are visible, an employment relationship is assumed to exist in the first place, and the burden of proving otherwise is left to the platform. According to one analysis, the presumption arises particularly when there is evidence that the working person receives instructions or is subject to the same control as employees belonging to the company, and that determination follows national laws, European Union case law, and collective agreements (DOI: 10.2139/ssrn.5706383, preprint; DOI: 10.52694/les.i/2025.1).

This directive did not close all the answers either. An analysis suggests that while the 2021 Commission draft had specific criteria triggering the presumption, they failed to reach a political compromise and were dropped in the 2024 adopted version, leaving only more abstract and ambiguous wording. Setting specific criteria was left to member states (DOI: 10.2139/ssrn.5386470, preprint). Amid intense lobbying by platforms and unions, member states were sharply divided, and the last-minute agreement passed without the support of France and Germany (DOI: 10.1111/jcms.70085). Member states must transpose the directive into national law by December 2026. A preprint study covering Germany, France, Spain, and Belgium highlights two uncertainties: how to handle intermediaries in subcontracting chains, and through what procedures the presumption should be asserted. Compliance has not been well achieved even in Spain and Belgium, which already have presumption rules, and the authors conclude that legal presumption alone is not enough without administrative pathways and active enforcement (DOI: 10.2139/ssrn.6867441, preprint).

Intermediaries in subcontracting chains are precisely the structure that South Korean court rulings have dissected one by one. South Korean courts traced back from the app operator through the staffing agency all the way to the car rental company. The European Union left that task to presumption rules and member state procedures, and those procedures are still being built.

The directive also carved out a separate path distinct from worker status. It is evaluated as having greatly expanded the scope of application of labor rights by granting rights regarding algorithmic management and collective rights even to platform workers who are genuine self-employed individuals (DOI: 10.2139/ssrn.4957467, preprint). The idea is to regulate how algorithms treat people regardless of whether they are workers.

What Has Not Been Settled Yet

First is the trade-off between flexibility and protection. One political philosophy paper views employment as a relationship where the working person gives up a portion of their freedom in exchange for income security. In this framework, the criticism of misclassification gains traction when platforms try to have it both ways. This is the case when workers give up their freedom but do not receive the corresponding security. The author argues that whether platforms provide more security or more freedom, they must do at least one (DOI: 10.1177/1470594x211015467). Viewed through this framework, the question to ask is whether the magnitude of the freedom surrendered matches the protection received.

Protection also comes with costs felt by workers. A preprint study surveying a representative sample of Uber and Rappi drivers and riders in Chile reports that Chile's two laws on platform work had minimal or negative impacts on workers. Regarding the first law, which formalized the status of platform workers as employees, those who cited negative impacts pointed to reduced scheduling flexibility and the obligation to pay income taxes. For the second law, which raised the requirements for becoming a driver, more than half of respondents said they would quit working. The study also revealed that the laws are only partially observed (DOI: 10.2139/ssrn.5285309, preprint, single country in Chile). Because it is a survey of one country at one point in time, it cannot be generalized to other countries. However, it shows that how laws increasing protection are experienced by working people must be measured separately.

The opposing argument is equally solid. A 2017 normative paper argues that labor law exists to address the inherent vulnerability of employment—specifically, the democratic deficit and economic and socio-psychological dependency created by broad subordination—and concludes on this basis that Uber drivers should be viewed as workers (DOI: 10.20318/sllerj.2017.3921). The drivers in the Tada case also claimed to be proud freelancers, but the lower court stated that such statements did not significantly influence its judgment, and the Supreme Court cited penalty clauses in training materials and notices of disadvantages in performance evaluations on the app screen as evidence of control.

Second is whether to establish an intermediate status. One legal study borrows the International Labour Organization's classification to place platform workers in a gray area between employees and independent contractors, and lists three paths to protection: broadening the concept of employees, setting a third category, and enacting separate legislation (DOI: 10.62795/fjl.v4i1.259). The United Kingdom already has an intermediate status called a worker. However, the author of the student-edited London School of Economics Law Review article reviewed earlier believes that the British system itself is sound and the problem lies in the courts' formalistic application. They argue that reviving a teleological interpretation centered on economic control and dependency can solve the issue without creating a fourth category (DOI: 10.61315/lselr.1047). This is where the divide lies on whether an intermediate status is a bridge expanding protection or an exit hatch reducing worker status.

Third is how to supervise algorithmic management. A preprint paper dealing with US employment law argues that existing laws presuppose an identifiable decision-maker, accessible reasons, and contestable procedures, but algorithmic management severs the link between decisions and accountability. The author proposes enforceable transparency obligations, meaningful human oversight, and limits on full automation in significant employment decisions (DOI: 10.2139/ssrn.6670340, preprint). The comparative law paper mentioned earlier also concludes that we should start with workers' right of access to evaluation and compensation criteria, and human review of major automated decisions such as account suspensions (DOI: 10.63108/vab.ldd.2.15). Tools are uneven even within the European Union. One critical paper evaluates the framework as fragmented, noting that the Artificial Intelligence Act provides broad but weak protection, while the Platform Work Directive provides strong but narrow rights (DOI: 10.1177/29768624251396248). Disputes have already reached the courts. As of April 2025, 22 rulings and 9 administrative decisions contesting or acknowledging that algorithmic management violated labor law or privacy have been compiled in Europe (DOI: 10.2139/ssrn.3982735, preprint report).

The South Korean Supreme Court ruling did not answer this third question. It merely read app records and penalties as evidence of control; what criteria algorithms must disclose and who should re-examine automated decisions were not issues addressed by the ruling.

The Next Time You Get a Gig Through an App

The facts examined by courts and legislatures in this article boil down broadly to four questions. Who distributes the work? Who sets the price? What happens if you refuse? Who can cut that person off? If an app is doing these things instead, that may be user control. The South Korean Supreme Court answered yes in the factual circumstances of one driver, the UK answered the same question with an intermediate status, and the South African labor court reached a different judgment on who the employer is.

Legislation has not closed the answer either. In California, a ballot measure aimed at forcing app-based workers to be treated as employees passed, and the European Union's presumption rules await implementation while leaving concrete criteria to member states. How much flexibility to give up and how much protection to receive, and who will supervise the invisible manager, are still unsettled matters. The story of other courtrooms where algorithms take charge of judgments about people can be read further in "When an Algorithm Sets the Sentence — Is AI Recidivism Prediction Fair?".

Judgments

• Supreme Court 2024Du32973 (2024.07.25) — Revocation of Remedy Decision for Unfair Dismissal — Upholding the lower court's ruling that a freelance driver for an app-based "vehicle rental service including driver placement" provided labor in a subordinate relationship and thus is a worker under the Labor Standards Act, and that the employer is the vehicle rental company (plaintiff), while dismissing the plaintiff's appeal (unanimous decision by 4 Justices). Established the legal principles that when determining platform workers, indicators of subordination must be applied by considering the characteristics of labor management by algorithms and multiple business participants, and that the exclusion period when adding a respondent is based on the time of the initial remedy application.
• Seoul High Court 2022Nu56601 (2023.12.21) — Revocation of Remedy Decision for Unfair Dismissal — The lower court of the above case. Revoking the Seoul Administrative Court 2020GuHap70229 judgment (2022.7.8) and dismissing the plaintiff's (vehicle rental company) claim for revocation of the review decision. Ruled that the National Labor Relations Commission's review decision recognizing the worker status of the driver, the employer status of the plaintiff, and the unfair dismissal nature of the headcount reduction notice is lawful.

Source Papers

• "2021/24 Supreme Court confirms that Uber drivers are 'workers' (UK)" (2021) — DOI: 10.5553/eelc/187791072021006002013 — The UK Supreme Court unanimously confirmed that Uber drivers are workers, not independent contractors, and considers them to be working from the moment they log into the app and are ready to work.
• "2017/10 Uber drivers found to be workers (UK)" (2017) — DOI: 10.5553/eelc/187791072017002001011 — The first-instance employment tribunal (reported in 2017 cases) ruled that Uber drivers are workers, making them eligible for basic protections such as minimum wage, paid leave, and whistleblowing protection. About 40,000 drivers in the UK, with Uber announcing an appeal.
• "Purposive Reading of Uber BV v Aslam and IWGB v CAC (Deliveroo): Fee-for-Performance or Fee-for-Access" (2026) — DOI: 10.61315/lselr.1047 — The UK's three-tier structure of employee, worker, and self-employed; criticism that the Uber and Deliveroo rulings treat substitution rights and multi-apping as decisive proof of autonomy, leaving a "legal statelessness"; the argument that the system is sound and can be resolved through purposive interpretation without a fourth category.
• "The employment status of food delivery riders in Europe and the UK: Self-employed or worker?" (2021) — DOI: 10.1177/1023263x211051833 — Discrepancies among courts, such as the French Court of Cassation recognizing an employment relationship while the UK High Court denied worker status to Deliveroo riders.
• "2022/19 Deliveroo riders and Uber drivers qualify as employees: modern employer's authority in platform work (NL)" (2022) — DOI: 10.5553/eelc/187791072022007002009 — Dutch courts in 2021 ruled Deliveroo riders and Uber drivers to be employees, with the Amsterdam District Court recognizing a "modern command structure."
• "2020/43 ECJ clarifies 'worker' status under EU law in gig economy ruling (UK)" (2020) — DOI: 10.5553/eelc/187791072020005003018 — The European Court of Justice: determining status is up to national courts, but Yodel delivery drivers appear correctly classified as self-employed given their discretion over acceptance, competitor work, substitution, and scheduling, with independence and subordination being the decisive factors.
• "The Uber Drivers Case: Rasier Operations BV v. E TU – The Supreme Court Decision (Part Two)" (2026) — DOI: 10.59403/126ekg4 — The New Zealand Supreme Court ruled that Uber drivers are employees under Section 6 of the Employment Relations Act, focusing on the intent of the parties while supporting it with criteria of control, integration, and the reality of the business.
• "Notes: Revisiting Uber Drivers as Employees — Comparing South African and UK Law" (2024) — DOI: 10.47348/ilj/v45/i1a2 — The South African labour court argued that the CCMA ruling viewing Uber drivers as employees of Uber SA was incorrect, comparing it with the UK Aslam judgment.
• "Addressing Platform Workers' Employment Misclassification: Legal Frameworks, Enforcement Strategies and the New Platform Work Directive" (2025) — DOI: 10.2139/ssrn.5706383 — (Preprint) In 2021, 28.3 million EU platform workers and 5.5 million under algorithmic control, with the Commission presuming misclassification; employment presumption and shifting the burden of proof; national courts increasingly looking at functional markers.
• "New European Union regulations concerning work via digital platforms – Directive (EU) 2024/2831…" (2025) — DOI: 10.52694/les.i/2025.1 — Employment relationship presumed if there are signs of direction and control, burden of proof rests on the platform, and specific forms are determined by member states.
• "Analysis of the criteria for establishing the existence of an employment relationship of platform workers under the Platform Work Directive -high expectations and limited outcome?" (2025) — DOI: 10.2139/ssrn.5386470 — (Preprint) The presumption criteria from the 2021 draft were dropped in the 2024 adopted version due to failed compromises, leaving only ambiguous phrasing delegated to member states.
• "Regulating Uncharted Waters: The EU Platform Work Directive Between Uncertainty, Lobbyism and Politicisation" (2026) — DOI: 10.1111/jcms.70085 — Adopted through a last-minute compromise without the support of France and Germany amidst intense lobbying and member state divisions (based on 32 policy documents and interviews).
• "A Prospective Analysis of the Transposition of the EU Directive on Platform Work" (2026) — DOI: 10.2139/ssrn.6867441 — (Preprint) Implementation deadline of December 2026, subcontracting intermediaries and the design of the presumption procedure as key uncertainties, inadequate compliance in Spain and Belgium, and presumption alone being insufficient.
• "The EU Platform Work Directive: What's New, What's Missing, What's Next?" (2024) — DOI: 10.2139/ssrn.4957467 — (Preprint) Expanding the scope by granting algorithmic management and data rights even to genuinely self-employed platform workers.
• "Economic security & the regulation of gig work in California: From AB5 to Proposition 22" (2021) — DOI: 10.1177/20319525211063111 — 10 years of legal and ethnographic research; self-organized workers were crucial in passing AB5, platform companies sponsored and succeeded in a ballot measure during the pandemic, criticized for anti-democratic implications.
• "Prop 22 and lessons for gig workers organising against algorithmic management" (2022) — DOI: 10.4324/9781003161875-23 — Prop 22, backed by Uber, Lyft, and others, passed with 58% approval on November 3, 2020, followed by concerns over labour-management deals cementing a third legal category.
• "Navigating the gig economy: transportation labor challenges facing California's app-based ridehailing and courier drivers" (2026) — DOI: 10.1016/j.tbs.2025.101218 — 41 literature sources, 8 experts, 7 case studies; drivers worry about fair pay, benefits, classification, and account deactivation transparency, with minimum pay and deactivation appeals proposed as solutions.
• "Algorithmic Management of Work on Online Labor Platforms: When Matching Meets Control" (2021) — DOI: 10.25300/misq/2021/15333 — Qualitative study of Uber drivers: algorithms have two main axes of matching and control, and drivers experience tensions regarding performance, pay, and affiliation.
• "Platform-Capital's 'App-etite' for Control: A Labour Process Analysis of Food-Delivery Work in Australia" (2019) — DOI: 10.1177/0950017019836911 — Qualitative case study of food delivery in Australia: control is multilayered, featuring surveillance-based and information asymmetries that constrain choices and lead to opaque performance management, with limited and individual responses.
• "Algorithmic Control in Platform Food Delivery Work" (2019) — DOI: 10.1177/2378023119870041 — 55 interviews and 955 surveys; all platforms assign and evaluate via algorithms, but with large variations, and Instacart acting as an "algorithmic dictatorship."
• "PLATFORM ÇALIŞANLARININ HUKUKİ STATÜSÜ: UBER-ASLAM VE DİĞERLERİ KARARINA İLİŞKİN BİR DEĞERLENDİRME" (2021) — DOI: 10.34246/ahbvuhfd.933999 — Summarizing the subordination criteria of the Aslam ruling, with the view that not all platform workers can be considered employees and each case must be judged individually.
• "The Invisible Employer: Algorithmic Management and the Reinvention of Control in the Gig Economy" (2026) — DOI: 10.63108/vab.ldd.2.15 — The argument that algorithmic management does not eliminate control but shifts and hides it, proposing human review for evaluation and pay access criteria, account deactivations, and more.
• "Governing Without Decision Makers Algorithmic Management and the Structural Failure of U.S. Employment Law" (2026) — DOI: 10.2139/ssrn.6670340 — (Preprint) Algorithmic management severs the connection between decisions and accountability, proposing transparency obligations, human oversight, and limits on fully automated decisions.
• "Regulating AI in the workplace: A critique of the EU AI act and the platform work directive through a worker-centred lens" (2025) — DOI: 10.1177/29768624251396248 — The AI Act is broad but weak while the platform directive is strong but narrow, resulting in a fragmented framework.
• "Case Law on Algorithmic Management at the Workplace: Cross-European Comparative Analysis and Tentative Conclusions" (2021) — DOI: 10.2139/ssrn.3982735 — Analysis of 22 court rulings and 9 administrative decisions related to algorithmic management in Europe as of April 2025.
• "On the (mis)classification of paid labor: When should gig workers have employee status?" (2021) — DOI: 10.1177/1470594x211015467 — Employment is a relationship of trading freedom for security, and criticisms of misclassification hold validity when platforms try to have it both ways, requiring them to provide at least one of security or freedom.
• "Impact of legislation on gig workers: evidence from Chile" (2025) — DOI: 10.2139/ssrn.5285309 — (Preprint, Chile) The impacts of the two laws are minimal to negative, pointing to reduced flexibility and income tax obligations, with over half responding they would quit if laws with stricter requirements passed, leading to partial compliance.
• "The Status of Uber Drivers: A Purposive Approach" (2017) — DOI: 10.20318/sllerj.2017.3921 — Normative argument that, in light of the purposes of labour law (democratic deficit and dependency arising from subordination), Uber drivers should be viewed as employees.
• "Gig Economy Worker's Legal Status: Employee or Independent Contractor?" (2024) — DOI: 10.62795/fjl.v4i1.259 — The grey area of dependent self-employment under ILO classifications, with three protective measures proposed: expanding the concept of worker, a third category, or independent legislation.