Deliveroo Italy algorithm linked to discriminatory rider shift allocations

An Italian court ruled in early January 2021 that an algorithm used by Deliveroo to rate riders and help allocate shifts was discriminatory. Subsequent reporting and Italian prosecutors' actions in February 2026 placed Deliveroo Italy under judicial supervision amid allegations that platform management and algorithmic shift rules contributed to unfair working conditions. Multiple press outlets and an AI incident repository document the ruling and the later supervisory measure.

Deliveroo Italy · Incident Jan 2, 2021 · Indexed Jun 10, 2026 · 5 sources

A rider-ranking and shift-allocation algorithm produced outcomes a court found discriminatory, reducing some riders' access to shifts and earnings.
What
An Italian court ruled in early January 2021 that an algorithm used by Deliveroo to rate riders and help allocate shifts was discriminatory.
Incident date
Jan 2, 2021
Who
Deliveroo Italy
Failure mode
Policy Violation
AI surface
Algorithmic Decision
Severity
High

What happened

In January 2021 an Italian tribunal found that an algorithm used by Deliveroo to evaluate riders and inform shift management was discriminatory. The ruling focused on how the platform's automated reliability/rating mechanisms affected riders' access to shifts and work. In February 2026 Italian prosecutors placed Deliveroo Italy under judicial supervision amid allegations that platform practices, including algorithmic management, had contributed to labour exploitation and inadequate pay and conditions for riders.

What broke inside the model

Failure path · mode profile · Policy Violation
  1. 01 · TriggerA prompt pushes against a deployment boundary.
  2. 02 · Model stepThe model produces the disallowed output.
  3. 03 · Control gapNo enforcement blocks it at generation time.
  4. 04 · FailureThe output crosses the policy line.
  5. 05 · ConsequenceA limit the business set is breached in public.

The output crosses a policy boundary the deployment had defined.

The failure centered on an automated scoring and scheduling system that ranked riders on reliability and used those rankings to prioritise shift allocation. The system relied on operational metrics such as acceptance rates and delivery performance in ways a court found produced discriminatory outcomes. The algorithmic criteria and opaque allocation rules meant some riders were systematically de-prioritised for shifts, lowering their opportunities to earn.

Public visibilityHigh
Regulatory exposureActive
Customer impactClass-wide
Financial impactUnknown
Time to disclosureMonths
  1. PressRider: l'algoritmo di Deliveroo è discriminatorio, la sentenza è storicadissapore.com
  2. PressMilano, controllo giudiziario per Deliveroo. «Caporalato, sfrutta i rider con l'algo­ritmo»milano.corriere.it
  3. PressL'algoritmo Frank: così le piattaforme di delivery gestiscono i riderrainews.it
  4. PressCourt rules against Deliveroo's rider algorithm, citing discriminationbusiness-humanrights.org
  5. PrimaryAIAAIC - Deliveroo Italy rider shift management algorithmaiaaic.org
Permalinkhttps://failureindex.ai/failures/deliveroo-italy-algorithm-linked-discriminatory-rider
CitationAI Failure Index. "Deliveroo Italy algorithm linked to discriminatory rider shift allocations" (FI-0498). Realm Labs. https://failureindex.ai/failures/deliveroo-italy-algorithm-linked-discriminatory-rider (indexed Jun 10, 2026).
Share cardA branded image of this record for posts and slides.

Data fields CC-BY 4.0, prose citation permitted. Incident ID FI-0498. Full dataset at /data.

Note from Realm Labs, the Index steward

How Realm fits

Controls for this failure mode
  • Prism
  • OmniGuard

This entry sits in the index's predictive wing: a system that scores, ranks, perceives, or steers rather than generates. Realm's runtime layer is built for the generative and agentic systems now moving into these same decision seats, where it watches a model's internal state and holds an unsupported claim or an unchecked action before it commits. The control gap on this record, an automated decision that reached people with no runtime check in front of it, is the same gap. The index keeps predictive failures on the record because the pattern carries straight into the systems shipping today.