Can Ride-Hailing Drivers Sue Platforms’ Algorithms?

A project reflection by Center for Digital Society at Universitas Gadjah Mada, supported by the Pulitzer Center through the South to South (S2S) AI Accountability CoLab microgrant.

What would it mean to sue an algorithm? The question sounds almost absurd. Algorithms cannot be summoned to court, represented, or held personally responsible. Yet for millions of ride-hailing drivers in Indonesia, algorithms increasingly determine some of the most consequential aspects of their working lives. Algorithms influence what orders drivers receive, how much they earn, which routes they take, how their performance is evaluated, and whether they can continue working on a platform at all.

So perhaps the question is not really whether drivers can sue algorithms. The real question is more uncomfortable: When an algorithm makes a decision, who exactly are we supposed to hold accountable?

This question was at the heart of our project, "Dissemination as Praxis: Empowering Litigators to Contest Algorithmic Asymmetries in Indonesia." Supported through the Pulitzer Center's South-to-South AI Accountability CoLab, the project brought together researchers, litigators, and representatives of driver organizations to think through what algorithmic accountability could look like in practice.

Our workshop in February 2026 was deliberately situated between research dissemination and legal organizing. We wanted to see what happens when research about algorithmic management is taken out of an academic report and placed in the hands of people who could actually challenge it on legal grounds.

The algorithm as a cultural problem

Much of the public discourse on AI accountability tends to begin from technical premises: Is the model biased? Is the training data representative? Is the system sufficiently transparent? These are necessary questions, and they have rightly shaped both scholarly and regulatory agendas. Yet our engagement with platform workers suggests that there is another register of inquiry that remains comparatively underexamined. One concerned not with the technical properties of algorithmic systems, but with the cultural meanings we project onto them.

The algorithm, in much of the public and even scholarly imagination, is figured as something objective, mathematical, and self-executing, a system that computes rather than decides. This framing carries consequences that extend well beyond semantics. When a driver receives fewer orders, when a fare shifts without explanation, or when an account is suspended, the event can be narrated as the neutral output of a technical process rather than as the outcome of deliberate organizational choices. 

This creates a peculiar form of power. A company can exercise control through an algorithm while appearing to retreat from the decision. The platform does not necessarily have to say, "We have decided that you should receive fewer orders." Instead, the decision appears to emerge from a system of ratings, data, optimization, demand, supply, and automated rules.

Our research found precisely these kinds of experiences. Some used terms such as akun gacor (rewarded accounts) and akun anyep (dry accounts) to describe accounts they perceived as receiving preferential treatment. 

The result is more than technological opacity. It is an accountability problem disguised as technical complexity.

Who is responsible when nobody appears to decide?

This became particularly visible when we examined the legal status of drivers. Ride-hailing platforms generally describe drivers as partners rather than employees. On paper, this appears to establish a relationship between independent parties. But the workshop revealed a contradiction between this legal language and the material experience of platform work.

Platforms possess extensive information about drivers: their locations, performance metrics, acceptance rates, trip histories, and behavioral patterns. Drivers, meanwhile, have little knowledge of how those same data points are translated into ratings, incentives, or sanctions. In practice, millions of workers are governed by algorithmic systems without meaningful rights to explanation or appeal. Drivers may be economically dependent on the platform while the platform determines the conditions under which they can work. Partnership contracts are often standardized and non-negotiable, while application updates can change the conditions of participation. Meanwhile, algorithmic systems influence order allocation, pricing, performance evaluation, and suspension. The lexical terminology "partner," then, does more than describe a contractual relationship. It produces a particular cultural and legal imagination of the worker.

The strange disappearance of the employer

There is an interesting paradox at the heart of platform capitalism.

The more sophisticated the platform becomes at managing workers through data, the easier it can become to make the employer disappear. The technology does not necessarily eliminate managerial power. It reconfigures how managerial power is experienced and represented.

Our legal problem documentation identified algorithmic opacity across several dimensions: dynamic pricing, order allocation, route optimization, automated suspension, and performance management. Drivers are subjected to consequential decisions while having limited access to the information needed to understand or challenge them. 

From "AI ethics" to the politics of remedy

This is where we found the question of litigation particularly useful. Legal action forces us to ask a different question from the usual AI ethics discourse. Instead of asking “Is the algorithm fair?”, we asked “Who has the obligation to make it fair, and what can happen if they don't?”

During the workshop, litigators identified multiple possible routes. These included data subject access requests, administrative litigation, judicial review, collective action, and advocacy around the Draft Platform Worker Law. Some of these mechanisms can potentially be pursued without first resolving whether drivers are legally employees. This matters because employment status itself is one of the central disputes.

The Personal Data Protection (PDP) Law, for instance, offers a possible entry point for challenging automated decisions because drivers may demand information about personal data used in decisions affecting them. The workshop therefore identified data access requests not merely as a privacy mechanism, but as a potential way of turning algorithmic opacity into evidence. This changes how we understand transparency. Transparency is not valuable simply because people deserve to "know how AI works." It matters because knowledge can become a resource for contestation.

In this sense, courts and regulatory bodies may become the arenas where AI governance in Indonesia is first negotiated. Rather than emerging from grand declarations about AI, accountability may begin with a driver's complaint about a suspended account, an unexplained drop in orders, or a missing fare. 

The Pulitzer Center-supported investigation "Grab Fares Surge Under Opaque Algorithm" by Karol Ilagan and Federico Acosta Rainis was particularly crucial because it gave us a concrete object around which these questions could be organized. Investigative journalism can perform a kind of epistemic intervention. It makes something that is normally difficult to see publicly legible.

In our workshop, the reporting became a bridge between technical systems and lived experience. We translated its findings into case briefs and discussion problems that could be taken up by legal practitioners. But the workshop also made clear where journalism ends and legal work begins. A story can reveal a pattern. Litigation must establish a claim. A driver can describe an unexplained suspension. A legal strategy must determine what evidence can establish the decision, the harm, and the platform's responsibility.

This is why we came to see journalism, research, and litigation not as competing forms of accountability, but as different stages of an accountability chain.

Accountability may begin with a driver's complaint

One of our clearest lessons was that a workshop is not itself accountability. It can create connections. It can translate research. It can identify legal pathways. It can make a previously abstract problem actionable. But it cannot substitute for the much slower work of organizing evidence, sustaining coalitions, supporting workers, and pursuing cases.

This was particularly apparent in the frustration expressed by driver representatives. Years of advocacy have not necessarily translated into corresponding legislative progress. The problem is therefore not simply a lack of awareness. It is the difficulty of sustaining collective action against institutions and corporations with significantly greater resources and power.

For us, this changed the meaning of "capacity building." Capacity is not simply teaching litigators what algorithmic management is. It means building the ability to collect evidence, identify responsible actors, formulate legal claims, protect affected workers, and remain organized long enough to pursue a remedy.

We are also exploring the development of a Litigation Hub that would bring researchers, journalists, lawyers, digital rights advocates, and workers together around actual or anonymized cases. The longer-term ambition is to identify a strategic case capable of establishing precedent around employment status, algorithmic transparency, or both. 

Perhaps, then, the real question is not whether we can sue an algorithm. It is whether we can refuse the idea that an algorithm is the end of the accountability chain. Because behind every supposedly autonomous system are institutions that design it, corporations that deploy it, regulators that permit it, workers whose data sustains it, and political choices that determine what kinds of harm are considered acceptable. The algorithm may not be able to stand in court. But the people and institutions behind it can.