Managed by Bots: Data Driven Exploitation in the Gig Economy

Managed by Bots: Data Driven Exploitation in the Gig Economy is the 13 December 2021 report of Worker Info Exchange (WIE), co-authored by WIE Research Lead Cansu Safak and Director James Farrar with a foreword by Bama Athreya, then a Fellow of the Open Society Foundations, and published as a joint civil-society campaign with the App Drivers and Couriers Union (ADCU), Privacy International, and Big Brother Watch, with named cooperation from the Equality and Human Rights Commission and Transport for London. The report is the first sustained civil-society documentation of how seven major UK and European gig platforms — Amazon Flex, Bolt, Deliveroo, Free Now, Just Eat, Ola, and Uber — use algorithmic systems to allocate work, set pay, score performance, run facial-recognition fraud-detection, and dismiss workers, and how those platforms routinely obstruct workers' attempts to exercise the data subject access rights granted under the EU General Data Protection Regulation that would be needed to assess fairness or contest decisions. Within the make-AI-good corpus Managed by Bots is the principal civil-society publication-side documentary artefact on algorithmic management of platform workers, the foundational documentary spine of the Worker Info Exchange / ADCU Amsterdam robo-firing and dynamic-pay litigation campaign, and the report whose framings — "robo-firing", "misclassification 2.0", management "behind the digital curtain" — anchor the corpus's standing vocabulary on platform-labour and algorithmic-management questions.

Origin and context

The report sits inside the multi-year UK platform-driver organising lineage that James Farrar and Yaseen Aslam built out of the Aslam and Farrar v Uber employment-status fight — the 28 October 2016 Central London Employment Tribunal ruling that Uber drivers were workers rather than independent contractors, affirmed unanimously by the UK Supreme Court in February 2021 — and its post-2016 successor question: once the employment-status fight had been won, the operating frontier of platform-labour organising moved to the question of how the platform's algorithms governed the working day and what the workforce could see of them. WIE was incorporated in October 2018 as the technical-legal vehicle for organising around that question and paired with the App Drivers and Couriers Union as the trade-union side of a data-trust model. Managed by Bots is the report through which WIE's data-trust model was operationalised at scale for the first time: the recovered subject access data from drivers across seven platforms was aggregated into a single civil-society evidentiary record, and the report made the substantive argument that that aggregated record warranted in the regulatory, judicial, and parliamentary registers.

Structure and content

The report has a six-part structure: a Summary of headline findings; an Introduction situating the report inside the post-employment-status platform-labour conversation; Part I — Misclassification 2.0: Controlled by Algorithms, which documents the algorithmic-management practices of the seven platforms and frames them as a successor exploitation pattern that conceals employment relationships even where workers have already secured statutory recognition; Part II — Exercising Data Rights at Work: Access, which documents the procedural obstacles drivers face in attempting to exercise their GDPR Article 15 subject access rights, including refusals to provide algorithm scoring information, guidance documents, or location data, and the broader pattern of incomplete or denied returns; Part III — Exercising Data Rights at Work: Litigation, which surveys the developing GDPR Article 22 case law on automated decision-making and frames the WIE/ADCU Amsterdam claim against Uber and Ola as the lead European test case; and a Conclusion that maps the report's findings onto the legal-and-regulatory architecture beginning to take shape around the then-proposed EU Directive on platform work, naming Article 6 of the directive proposal as the legal instrument under which the report's findings would translate into binding worker protections at European level.

Case studies — robo-firing in practice

The report's evidentiary spine is two named case studies that document the "robo-firing" pattern its substantive argument is built around. Pa Edrissa Manjang, a Black Uber driver in the UK, was wrongfully deactivated by an automated facial-recognition selfie-verification system that repeatedly failed to match his selfies against his profile photograph; the platform terminated his access without meaningful human review of the underlying mismatch. Aweso Mowlana, also an Uber driver, was suspended and dismissed after the platform's geolocation-monitoring system misinterpreted overlapping signals from his and his brother's accounts as evidence that he had been sharing his account in violation of platform terms; the algorithm-generated allegation was treated as substantively decisive and the dismissal followed without an evidentiary process Mowlana could have engaged. The two cases together establish what the report calls the structural pattern: an automated profiling system generates a high-stakes adverse decision about a worker's livelihood; the platform's human review either does not occur or amounts to a rubber-stamp; and the worker is denied access to the underlying data needed to contest the decision. Both cases became the documented examples around which the WIE/ADCU Amsterdam litigation was framed, and both became reference points in the broader European civil-society register on algorithmic management.

The "misclassification 2.0" framing

The report's central conceptual contribution is the misclassification 2.0 framing carried in Athreya's foreword and elaborated in Part I. The argument is that the first wave of platform-labour exploitation depended on the legal misclassification of drivers and couriers as independent contractors rather than workers, denying them statutory employment protections; once the employment-status fight had been substantially won — through the UK Supreme Court ruling in Aslam and Farrar v Uber and comparable rulings in other European jurisdictions — the platforms' exploitation register shifted into a second wave that depends on algorithmic control of the working day rather than on the contractual classification of the employment relationship. As Farrar framed the argument at the report's release, "the report shows how the latest wave of employment misclassification tactics involves employers telling workers they are truly independent in their jobs, while at the same time management control is wielded as forcefully as ever but from behind the digital curtain". Athreya's foreword pairs the framing with the substantive accusation that "platform companies are operating in a lawless space where they believe they can make the rules" — naming the regulatory void in which algorithmic management has been built as the structural pre-condition that European data-protection law, properly used, can begin to close. The framing has since anchored the corpus's standing vocabulary on algorithmic management as the successor question to the employment-classification fight.

Documented platform practices

Across the seven platforms the report documents a common pattern: algorithmic work allocation and prioritisation that workers cannot see or contest; extensive driver-monitoring data collection through GPS, facial-recognition, in-app surveillance, and behavioural inference; automated or semi-automated dismissal and pay-reduction decisions; facial-recognition identity-verification systems used as a deactivation surface; and "woefully inadequate levels of transparency" regarding the underlying decision-making logic. The report's data-access findings document that platforms routinely deny workers' Article 15 subject access requests outright, return incomplete data sets, withhold algorithm-scoring information, refuse to disclose internal guidance documents, and decline to release location-data returns — and that when human review of automated decisions does occur, the platforms' implementations function as a "rubber-stamp" rather than the meaningful intervention GDPR Article 22 contemplates. Cansu Safak's day-of-release statement that "gig platforms are collecting an unprecedented amount of data from workers through invasive surveillance technologies" names the data-collection asymmetry as the structural condition under which the algorithmic management register operates. ADCU President Yaseen Aslam's paired statement that "our union has been overwhelmed with casework from workers who were summarily dismissed by gig economy bosses after unsubstantiated allegations" names the on-the-ground operational caseload through which the substantive findings became apparent to the trade-union side of the partnership.

Coalition, campaign, and reception

The report was published as the lead documentary artefact of a joint Managed by Bots public campaign with Privacy International, the App Drivers and Couriers Union, and Big Brother Watch, with named cooperation from the Equality and Human Rights Commission and Transport for London and with legal representation from Bates Wells. The campaign's central public-facing ask was for binding employer transparency, accountability, and redress obligations in respect of algorithmic management — translated through the report's specific findings into demands for platform compliance with Article 15 subject access requirements, meaningful Article 22 human-review protections against automated dismissal, and the inclusion of binding algorithmic-management protections in the then-proposed EU Directive on platform work. The report's UK and European specialist-press reception was carried in particular by Sebastian Klovig Skelton's Computer Weekly coverage on 15 December 2021, which centred Farrar's "behind the digital curtain" framing as the conceptual headline of the report. International-civil-society uptake travelled through Privacy International's long-read documentation of the campaign and through the Open Society Foundations editorial circuit out of which Athreya's foreword had emerged, including her prior Inequality.org long-form framing of the platform-labour data-rights argument.

Travel into litigation

The report's most consequential downstream artefact is its evidentiary carry into the WIE/ADCU Amsterdam litigation, which had been filed in 2020 and was on appeal at the time of the report's release. The Manjang and Mowlana case studies and the broader Article-15-access pattern the report documented became the substantive evidentiary base on which the Amsterdam Court of Appeal's three linked judgments on 4 April 2023 (ECLI:NL:GHAMS:2023:793, 796, and 804) found that Uber and Ola's ride assignment, dynamic pricing, driver rating, fraud-probability scoring, and account deactivation processes qualify as automated decision-making within the meaning of GDPR Article 22, that the platforms' "limited human review" amounted to little more than "a purely symbolic act", and that the platforms could not invoke trade-secrets exemptions to withhold information about how the underlying algorithms work. The report is named in the WIE litigation archive as the documentary anchor of the underlying claim. The substantive line that begins in the report's two case studies has continued through the 5 October 2023 €584,000 non-compliance penalty against Uber and into the 20 November 2025 Stichting WIE International collective action against Uber's algorithmic pay-setting system — making Managed by Bots the publication-side foundation of the longest-running European civil-society litigation arc on algorithmic management to date.

Position within the corpus

Within the corpus Managed by Bots is the first sustained civil-society publication-side documentary artefact on algorithmic management of platform workers, the foundational documentary spine of the Worker Info Exchange / ADCU Amsterdam robo-firing and dynamic-pay litigation campaign, and the report that anchors the corpus's standing vocabulary on platform-labour algorithmic management — "robo-firing", "misclassification 2.0", management "behind the digital curtain", the broader algorithmic-management framing the corpus's worker-organising entries operate from. The report closes the corpus's depth gap on the publication side of the gig-worker / algorithmic-management organising lineage: WIE was already in the corpus as an organisation with its own body and the Amsterdam litigation as a campaign; the report was load-bearing in both entries' bodies without being a standalone publication entry of its own, and is also body-named in the voice-james-farrar, person-james-farrar, msg-algorithmic-management, and msg-fairwork entries.

As a publication type Managed by Bots fills the report slot in the worker-organising publication slate, sitting alongside the African Content Moderators Union and Foxglove Kenyan content-moderator litigation campaigns on the African continent: where the Kenyan litigation has tested whether US-headquartered platforms can be held accountable in Kenyan courts on behalf of African workers performing the labour that trains and supports their products, Managed by Bots is the publication-side artefact that tested whether European data-protection law operates as a worker-power instrument against the algorithmic management of European gig drivers' working lives. Among the corpus's existing algorithmic-public-administration reports — the La Quadrature du Net CAF report on the French family-benefits agency's beneficiary-scoring algorithm, the AlgorithmWatch Automating Society report on European public-sector ADM mapping, the Panoptykon Foundation Black-Boxed Politics essay, and the Derechos Digitales Latin American ADM mapping — Managed by Bots is the equivalent foundational documentary artefact on the private-sector employer-employee side of automated-decision-making contestation, completing the public-administration / private-employer pair on the corpus's European civil-society publications slate. The report's lasting movement-side contribution is that it operationalised the data-trust model as a documented civil-society research method: the recovered subject access data from drivers across seven platforms was the substantive evidentiary base on which the report's findings, the misclassification 2.0 framing, and the subsequent multi-year litigation arc were built, and the methodology is now the corpus's clearest demonstration of how individual data subject access rights granted under European data-protection law can be aggregated by a civil-society organisation into a collective evidentiary infrastructure for worker-side challenge to algorithmic management.