In recent years, app-based platforms have brought major changes to the way people work. Ride-hailing, food-delivery and household-services apps now provide work to a large number of people, and for those without other employment these platforms act as a last resort. According to a 2022 NITI Aayog report, India had around 7.7 million gig workers in 2020–21, a number expected to grow to 23.5 million by 2029–30, when gig workers would form 6.7 per cent of the non-agricultural workforce (4.1 per cent of total livelihood) in the country.1 The COVID-19 pandemic accelerated the shift to digital services, and platform work held up during it while several other sectors of the economy were adversely affected.2 Companies such as Uber, Ola, Zomato, Swiggy and Urban Company now depend on a large number of flexible workers who are managed through apps. This type of work is often called “flexible work”. However, it raises an important legal question: what is the actual relationship between the platform and the worker? The answer decides whether the worker can claim minimum wages, social security, the right to bargain collectively and basic protection against penalties or removal from the platform.
Indian labour law recognises two kinds of work arrangement. The first is a contract of service, which creates an employer-employee relationship and gives the employee several legal protections. The second is a contract for service, under which the person is treated as an independent contractor, and to whom most labour laws do not apply. Gig workers are currently placed in the second category. In reality, however, many gig workers depend on a single platform for their entire income and livelihood. The platform decides how they should work and how much they are paid, and closely monitors their performance. In some cases, a worker’s account can even be blocked without prior warning or without a proper chance to explain their side. The main issue examined in this paper is that gig workers do not clearly fit into either of the existing categories; as a result, they often have no clear legal right to minimum wages under the Code on Wages, 2019 or to collective bargaining under the Industrial Relations Code, 2020. There is also a lack of proper safety protection at work and of clear protection against sexual harassment at the workplace.
Research question: What are the legal problems in deciding whether a gig worker is an employee, and how can these problems be solved?
Objectives of the study:
1. To find out why the traditional tests used by the courts to decide employment status do not work well when an algorithm takes charge.
2. To understand how the Code on Social Security, 2020 deals with gig and platform workers, and to examine the difference between providing them with welfare benefits and giving them full employee status.
3. To compare how the United Kingdom, California and the European Union have addressed the same issue.
4. To suggest a workable legal solution for India without affecting the flexibility that makes this kind of work attractive.
Scope of the study: This paper focuses mainly on Indian labour law. The approaches followed in the United Kingdom, California and the European Union are included only for comparison, each jurisdiction having dealt with the classification of gig workers in its own way. Within India, the study concentrates on transport and delivery platforms, since issues of algorithmic control arise, and have been challenged before the courts, chiefly in these sectors.
This paper follows a doctrinal and comparative research method. It examines statutes and court judgments as primary sources, and uses government reports and academic writing as secondary sources. The comparison with other jurisdictions is limited to those where the issue has already been addressed through legislation or significant court decisions. The study does not collect new field data; statistics such as those of NITI Aayog are used only to show the scale of the problem.
Indian courts generally follow three main tests to decide whether a person is an employee or an independent contractor. These tests were developed long ago, when human supervision was the norm. Understanding them helps to show their weaknesses when applied to gig workers, as the following part demonstrates.
A contract of service creates an employer-employee relationship and gives the worker legal protection under laws such as the Industrial Disputes Act, 1947 and the Employees’ State Insurance Act, 1948 (now subsumed in the Industrial Relations Code, 2020 and the Code on Social Security, 2020, which took effect on 21 November 2025), and the Code on Wages, 2019.3 A contract for service creates a relationship between a client and an independent contractor, to which most of these protections do not apply. The problem is that none of these laws clearly explains what makes a relationship a “contract of service” rather than a “contract for service”. Because of this gap, Indian courts developed the three tests set out below to decide whether an employer-employee relationship exists. The tests are often applied together rather than one strictly after another.
The control test is the oldest of the three. It asks whether the employer has the right to control not only what work is done but also how it is done. In Dharangadhara Chemical Works Ltd. v. State of Saurashtra,4 the Supreme Court held that the prima facie test of the relationship of master and servant is the existence of the master’s right to supervise and control the work, not only in directing what work the servant is to do but also the manner in which it is to be done. The Court also cautioned that the nature and extent of the control required must vary from business to business and is incapable of precise definition, noting English authority that the control test is not of universal application. On that approach, it upheld the finding that the agarias, salt workers paid by the piece, were workmen.
Since the control test does not work well in some cases, courts developed the integration test, which asks whether the worker’s job is an integral part of the employer’s business or merely an accessory service. In Silver Jubilee Tailoring House v. Chief Inspector of Shops and Establishments,5 the tailors were paid by the piece and were free to come and go as they chose. The Supreme Court observed that it is in skilled and particularly professional work that the control test in its traditional form has broken down, and referred to Denning L.J.’s view that being a servant depends on whether the person is “part and parcel of the organisation”. It held the tailors to be employees, because they worked on the proprietor’s premises with sewing machines that belonged to him, and because his right to reject work that did not follow his instructions showed the necessary element of control and supervision.
The economic reality test is the most comprehensive and workable of the three. Sometimes called the multiple-factor test, it looks at the entire relationship instead of a single factor: who owns the tools, who bears the financial risk, and how far the worker depends on the employer for income. In Hussainbhai v. Alath Factory Thozhilali Union,6 Krishna Iyer J. held that where workers produce goods or services for the business of another, that other is in fact the employer, because he has economic control over the workers’ subsistence, skill and continued employment, whatever the paper arrangement through an intermediate contractor may say. The Supreme Court revisited the question in Balwant Rai Saluja v. Air India Ltd.,7 where it listed the factors to be weighed (who appoints the workers, who pays them, who can dismiss and discipline them, whether there is continuity of service, and the extent of control and supervision) and held that canteen workers engaged through a subsidiary were not employees of Air India because they were not under its “effective and absolute control”. The two decisions show that the courts look past the labels in a contract to the working relationship actually in place, though Balwant Rai Saluja sets a demanding threshold of control.
All three tests rest on a similar assumption: that control, integration or dependence can be seen through direct human supervision, such as a manager giving instructions, a supervisor fixing working hours or a foreman directing how the work should be done. Platform work breaks this assumption completely. On a platform, control is exercised by software following rules that the platform itself has set, not by a person physically present. None of the three tests was developed for this situation, and Indian courts have not yet settled how they apply to platform work. The next part examines this gap: how the tests operate when supervision is carried out through algorithmic management instead of the direct human management they were designed to detect.
In the platform or gig economy, supervision is carried out not by a human manager but by software. This is called algorithmic management. The algorithm decides which task goes to which worker, changes prices on its own and monitors the worker’s performance continuously. It can also penalise a worker or block their account on the basis of ratings or acceptance rates. Often no human is involved in the decision and no proper explanation is given. In some ways this control is stronger than traditional human supervision: it works without pause, applies automatically and uniformly to every worker, and leaves almost no room for the worker to negotiate.
Platforms say that their workers are independent contractors because they can log in or log out whenever they choose. In practice, this freedom is narrower than it sounds. Incentive rewards push workers towards working during busy hours, and a worker who turns down tasks or logs off during periods of high demand may be penalised by the algorithm, for example by being offered fewer tasks in future. The worker’s choice to log off is real, but the platform alone decides the consequences of that choice. The worker’s freedom therefore exists only within limits that the platform itself controls.
When the three traditional tests are applied to platform work, each runs into difficulty. Under the control test, an algorithm that fixes the fare, decides the route and penalises a worker for declining tasks is arguably exercising exactly the kind of control the test was meant to capture, but no Indian court has yet finally decided the question on such facts. Under the integration test, platforms argue that they are neutral technology companies rather than transport or delivery providers. This claim is hard to accept, since the platform would have no business at all without its workers. Under the economic reality test, the picture is mixed. A gig worker may own the vehicle, which suggests independence; but the same worker may depend entirely on one platform for income and have no power to set a price, which suggests dependence. The test was never designed for factors that are constantly fixed and changed by an algorithm.
The Code on Social Security, 2020 is the first Indian law to define “gig worker” and “platform worker” as separate categories.8 However, the Code defines them only negatively, as persons who work outside a traditional employer-employee relationship, and does not give them a distinct set of positive legal rights. Chapter IX of the Code provides mainly for welfare schemes, financed in part through a Social Security Fund to which aggregators are required to contribute.9 The other three Labour Codes do not follow even this welfare-based approach. The Code on Wages, 2019 guarantees the minimum wage only to an “employee” employed on wages by an establishment, which leaves self-employed persons outside minimum-wage protection.10 The Industrial Relations Code, 2020 builds its collective-bargaining rules around a “worker” employed in an “industry”, which leaves gig workers out.11 The Occupational Safety, Health and Working Conditions Code, 2020 does not cover them either: its protections attach to “employees” and “workers” of an establishment, and it makes no mention of gig or platform workers.12 In the end, gig workers may obtain access to a welfare fund, but they still have no minimum wage, no right to bargain collectively and no proper safety protection at work.
Indian courts have still not settled the classification question. The most important pending case is Indian Federation of App-based Transport Workers v. Union of India,13 a writ petition asking the Supreme Court to declare that gig and platform workers are “unorganised workers” or “wage workers” and are entitled to social security. The petition argues that excluding them from these protections violates Articles 14, 21 and 23 of the Constitution. At the time of writing, the case remains pending before the Supreme Court, and no final decision has been reported.
In Ms X v. Internal Complaints Committee, ANI Technologies Pvt. Ltd.,14 the Ola case, the Karnataka High Court held that the degree of control Ola exercised over its drivers was sufficient to make a driver an “employee” within the meaning of the POSH Act, 2013, so that Ola’s Internal Complaints Committee was bound to inquire into a passenger’s complaint of sexual harassment by a driver. The Court rejected Ola’s argument that its drivers were independent contractors outside the Act. The ruling is under appeal before a Division Bench, which has stayed it.15 The decision is therefore not final, but it remains the clearest example so far of an Indian court directly examining the control question with which this paper is concerned.
The Code on Social Security, 2020 provides for a National Social Security Board, which also serves as the board for the welfare of gig and platform workers, and requires these workers to register by an application that includes their Aadhaar number.16 This is a useful step forward, but the Code deals only with welfare. It does not create an employer-employee relationship, and it leaves the actual classification question to be worked out by the courts using the tests discussed above.
Even the limited protections given by the Code are hard to put into practice. Labour authorities may well lack the technical expertise needed to examine platform algorithms. Labour is on the Concurrent List,17 so the central and state governments must coordinate, and progress has been slow: the Codes, enacted in 2019 and 2020, took effect only on 21 November 2025, and even then the rules and schemes under them had still to be framed.18 The Central Government also has the power to exempt aggregators from paying into the welfare fund, which could lead to uneven enforcement.19 On top of this, platform work sits, as NITI Aayog puts it, “at the unique intersection of formal-informal classification”, with formal characteristics (payments and credit) as well as informal ones (time management and dynamic work relations), while access to social security has traditionally turned on that classification.20 Taken together, these problems leave a large gap between what the Code promises and what actually happens.
Some progress has been made at the state level. The Karnataka Platform Based Gig Workers (Social Security and Welfare) Act, 2025 does not itself mention sexual harassment, but the Rules made under it require that all complaints of sexual harassment faced by women be dealt with in accordance with the POSH Act, 2013.21 In Karnataka, this addresses, through subordinate legislation rather than litigation, the gap in protection against sexual harassment identified earlier in this paper, although the Rules do not say which body is to inquire into such complaints. The Act also requires platforms to tell workers how to seek information about the automated monitoring and decision-making systems that affect their working conditions, and forbids termination or deactivation without valid reasons in writing and, save in cases involving bodily harm, fourteen days’ prior notice.22 Implementation of the Act has, however, been slow and contested: the State constituted the welfare board under the Act only in January 2026,23 and several platforms have since challenged the Act before the Karnataka High Court, which has declined to stay it but has granted them interim protection on condition that they deposit the welfare fee.24 The Code on Social Security, 2020 contains no comparable provision on sexual harassment.
In Uber BV v. Aslam,25 the UK Supreme Court held that Uber drivers are “workers”, a statutory middle category in UK employment law that sits between an employee and a genuinely self-employed contractor. This status entitles the drivers to the national minimum wage and paid annual leave. The Court declined to treat Uber’s written contracts, which described the drivers as independent, as decisive, and held that the tribunal was right to look at how the relationship actually operated in practice.
The Court relied on several facts:
1. Uber, not the driver, fixed the fare.
2. Uber dictated the contractual terms, with no room for the driver to negotiate.
3. A driver whose rate of accepting or cancelling trips fell outside the levels Uber set could be automatically logged off the app as a penalty.
4. Uber restricted communication between driver and passenger so that drivers could not build up their own client base.
Taking these factors together, the Court concluded that the drivers were in a position of subordination and dependence in relation to Uber. This is exactly the situation that worker-protection laws are meant to cover, whatever label the contract uses.
In Dynamex Operations West, Inc. v. Superior Court,26 the California Supreme Court adopted the “ABC test”. Under this test, a worker is presumed to be an employee unless the hiring entity proves all three of the following: that the worker is free from its control and direction, both under the contract and in fact; that the work is outside the usual course of its business; and that the worker is customarily engaged in an independently established trade or business of the same kind. The test was written into statute by Assembly Bill 5 in 2019. Then, after Uber, Lyft and other app-based companies reportedly spent some USD 200 million campaigning for it,27 California voters approved Proposition 22 in November 2020, which classifies app-based transport and delivery drivers as independent contractors when specified conditions are met, taking them outside the ABC test. Although a trial court held Proposition 22 unconstitutional in 2021, the Court of Appeal largely reversed that ruling, and in 2024 the Supreme Court of California upheld the measure in Castellanos v. State of California.28 California shows that even a strong presumption of employment, adopted by the courts and enacted by the legislature, can be undone through a well-funded political campaign and a popular vote.
The European Union’s Platform Work Directive was proposed by the European Commission in December 202129 and adopted in October 2024.30 The Directive addresses false self-employment among platform workers through a rebuttable legal presumption of employment: where facts indicating direction and control are found, the relationship between a digital labour platform and a person working through it is presumed to be an employment relationship, and it is for the platform to prove otherwise. The Directive also contains important rules on algorithmic management. Platform workers have the right to know when automated systems are used to monitor them or to take decisions about them; any decision to restrict, suspend or terminate a worker’s account must be taken by a human being; and workers may obtain an explanation of, and request a review of, decisions taken or supported by automated systems. Overall, the Directive is an important step towards making platform work fairer, more transparent and more protective of workers’ rights.
Three lessons emerge from this comparison. First, the United Kingdom shows that courts, by interpreting an existing statutory category purposively, can extend protection to platform workers without a new law. Second, California shows that a strong legal presumption of employment can be reversed by ordinary political pressure if it is not adequately entrenched. Third, the EU model shows that the most durable solution addresses the presumption of employment and algorithmic transparency together, within one strong law. These three lessons, namely flexibility through the courts, stronger protection through legislation, and the combination of presumption with transparency, form the basis of the recommendations that follow.
This paper recommends that legislation should create a rebuttable presumption of employment for platform workers. A person doing platform work under algorithmic control would then normally be treated as an employee for the purposes of wages, social security and collective bargaining, unless the platform can prove that the worker sets their own prices, can refuse work without penalty and is free to find clients outside the platform. The experience of Proposition 22 in California suggests that this rule should be placed in principal legislation, such as an amendment to the Code on Social Security, 2020, rather than in rules that can be changed more easily. It should also be recognised that platforms may respond by redesigning or tightening their algorithms to avoid the presumption, which is why the next two recommendations are equally important and should not be treated as optional.
The law, or the courts through interpretation, should clearly recognise that automated task allocation, rating-based penalties, unilateral pricing and automatic deactivation can all amount to control under the control test discussed above. If this is recognised, courts can apply existing case law, such as Dharangadhara Chemical Works,31 to platform work without waiting for an entirely new legal framework.
This paper also recommends giving workers a legal right to algorithmic transparency. Platforms should be required to explain clearly, in simple language, how they make decisions about task allocation, payment and deactivation. Workers should also have a right to fair human review: before a worker is deactivated, a real person should give a clear written reason for the decision, and the worker should have a chance to explain their side. This would also ease an evidential problem, since at present a worker who wishes to challenge an algorithmic decision may be unable to see or question the data on which it was based, because that information is held only by the platform.
Gig and platform work has grown very quickly in India, but the law has not kept pace. The legal system still divides workers mainly into employees and independent contractors, and this simple division does not work well for gig workers. The Code on Social Security, 2020 is an important step, but it provides only a partial solution. The traditional tests of control, integration and economic reality have also not been properly tested against algorithmic management. The experience of the United Kingdom, California and the European Union suggests a way forward for India: a legal presumption of employment, clear recognition of algorithmic control, and enforceable rights to transparency and human review. Together, these steps can provide a practical solution, one that protects platform workers properly without taking away the flexibility that makes this type of work attractive to so many people.
*****
1. NITI Aayog, India’s Booming Gig and Platform Economy: Perspectives and Recommendations on the Future of Work xvii, 24 (June 2022), https://www.niti.gov.in/sites/default/files/2022-06/25th_June_Final_Report_27062022.pdf.
2. NITI Aayog, supra note 1, at 4, 29.
3. The Industrial Relations Code, 2020, No. 35, Acts of Parliament, 2020, § 104 (India) (repealing the Industrial Disputes Act, 1947); The Code on Social Security, 2020, No. 36, Acts of Parliament, 2020, § 164 (India) (repealing the Employees’ State Insurance Act, 1948); The Code on Wages, 2019, No. 29, Acts of Parliament, 2019 (India); Press Release, Ministry of Labour & Employment, Government Makes the Four Labour Codes Effective to Simplify and Streamline Labour Laws (Nov. 21, 2025), https://www.pib.gov.in/PressReleasePage.aspx?PRID=2192463.
4. Dharangadhara Chemical Works Ltd. v. State of Saurashtra, AIR 1957 SC 264, 1957 SCR 152, 160 (India).
5. Silver Jubilee Tailoring House v. Chief Inspector of Shops & Establishments, (1974) 3 SCC 498 (India).
6. Hussainbhai v. Alath Factory Thozhilali Union, (1978) 4 SCC 257 (India).
7. Balwant Rai Saluja v. Air India Ltd., (2014) 9 SCC 407, ¶¶ 65, 88 (India).
8. The Code on Social Security, 2020, No. 36, Acts of Parliament, 2020, § 2(35) (“gig worker”), § 2(60) (“platform work”), § 2(61) (“platform worker”) (India).
9. Id. §§ 109–114 (ch. IX), 114(4), 141.
10. The Code on Wages, 2019, No. 29, Acts of Parliament, 2019, §§ 2(k), 5 (India).
11. The Industrial Relations Code, 2020, No. 35, Acts of Parliament, 2020, § 2(p) (“industry”), § 2(zr) (“worker”) (India).
12. The Occupational Safety, Health and Working Conditions Code, 2020, No. 37, Acts of Parliament, 2020, § 2(1)(t) (“employee”), § 2(1)(zzl) (“worker”) (India).
13. Indian Federation of App-based Transport Workers v. Union of India, W.P. (C) No. 1068 of 2021 (India) (pending); see Gig Workers’ Access to Social Security, Supreme Court Observer, https://www.scobserver.in/cases/gig-workers-access-to-social-security-the-indian-federation-of-app-based-transport-workers-ifat-v-union-of-india/ (last visited Oct. 4, 2026) (listing the case as pending after the hearing of Feb. 18, 2025).
14. Ms X v. Internal Complaints Committee, ANI Technologies Pvt. Ltd., W.P. No. 8127 of 2019 (GM-RES) (Kar. HC Sept. 30, 2024) (India).
15. Salil Tiwari, Karnataka HC Stays Order Directing Ola to Pay Rs 5.5 Lakh in Passenger’s Sexual Harassment Case, LawBeat (Oct. 9, 2024), https://lawbeat.in/news-updates/amp/karnataka-high-court-stays-order-directing-ola-pay-rs-55-lakh-passengers-sexual-harassment (reporting the Division Bench’s interim order of Oct. 4, 2024 in ANI Technologies Pvt. Ltd. v. Ms X); see also Lokesh Choudhary, Relief for Ola: Karnataka HC Stays Order Declaring Drivers as Startup’s Employees, Inc42 (Oct. 5, 2024), https://inc42.com/buzz/relief-for-ola-karnataka-hc-stays-order-declaring-drivers-as-startups-employees/.
16. The Code on Social Security, 2020, No. 36, Acts of Parliament, 2020, §§ 6(1), 113(2), 114(6) (India).
17. India Const. sched. VII, list III, entries 22–24.
18. Press Release, Ministry of Labour & Employment, supra note 3.
19. The Code on Social Security, 2020, No. 36, Acts of Parliament, 2020, § 114(7)(ii) (India).
20. NITI Aayog, supra note 1, at 73.
21. The Karnataka Platform Based Gig Workers (Social Security and Welfare) Act, 2025, Karnataka Act No. 72 of 2025 (India); The Karnataka Platform Based Gig Workers (Social Security and Welfare) Rules, 2025, r. 12(6), Notification No. LD 413 LET 2023 (Nov. 19, 2025) (India); Vinay Joy, Srishti Ramkrishnan & Ajay Kranthi Kothwal, Karnataka Platform Based Gig Workers (Social Security and Welfare) Act 2025, Mondaq (Dec. 3, 2025), https://www.mondaq.com/india/employee-rights-labour-relations/1713454/karnataka-platform-based-gig-workers-social-security-and-welfare-act-2025.
22. The Karnataka Platform Based Gig Workers (Social Security and Welfare) Act, 2025, supra note 21, §§ 13(1), 14(2).
23. Aakriti Bansal, Karnataka Govt Constitutes Gig Workers Welfare Board: What Comes Next, MediaNama (Jan. 29, 2026), https://www.medianama.com/2026/01/223-karnataka-govt-gig-workers-welfare-board/.
24. Siddesh M S, Gig Workers Act: Karnataka HC Grants Interim Protection to IAMAI, Swiggy, Zomato, Others but Orders Deposit of Welfare Fee, Bar & Bench (July 3, 2026), https://www.barandbench.com/news/gig-workers-act-karnataka-hc-grants-interim-protection-to-iamai-swiggy-zomato-others-but-orders-deposit-of-welfare-fee; Rohit Singh, Uber Challenges Karnataka’s Gig Workers Act in High Court, Gets Interim Relief, MediaNama (July 30, 2026), https://www.medianama.com/2026/07/223-uber-challenges-karnatakas-gig-workers-act-interim-relief/.
25. Uber BV v. Aslam [2021] UKSC 5, ¶¶ 1, 34, 94–101 (appeal taken from Eng.).
26. Dynamex Operations West, Inc. v. Superior Court, 4 Cal. 5th 903 (2018), codified by Assemb. B. 5, 2019–2020 Reg. Sess. (Cal. 2019), and substantially carved back for app-based transportation and delivery drivers by Proposition 22 (Cal. 2020), Cal. Bus. & Prof. Code § 7451.
27. Brian Melley, Uber, Lyft Spend Big, Win in California Vote About Drivers, PBS NewsHour (Associated Press, Nov. 4, 2020), https://www.pbs.org/newshour/politics/uber-lyft-spend-big-win-in-california-vote-about-drivers.
28. Castellanos v. State of California, 16 Cal. 5th 588 (2024), aff’g 89 Cal. App. 5th 131 (2023).
29. Proposal for a Directive of the European Parliament and of the Council on Improving Working Conditions in Platform Work, COM (2021) 762 final (Dec. 9, 2021).
30. Directive (EU) 2024/2831 of the European Parliament and of the Council of 23 October 2024 on Improving Working Conditions in Platform Work, arts. 5(1)–(2), 9(1), 10(5), 11(1)–(2), 2024 O.J. (L 2024/2831).
31. Dharangadhara Chemical Works, supra note 4, 1957 SCR at 160.