Uber Eats courier alleges racial bias after facial-verification mismatches and dismissal
A UK Uber Eats courier, Pa Edrissa Manjang, alleges he faced excessive facial-photo verification checks and was deactivated from the app in April 2021 after repeated mismatches. He brought a discrimination claim that a tribunal allowed to proceed and later received a payout, while Uber has said automated facial verification was not the reason for the temporary loss of access.
Automated facial-verification mismatches repeatedly locked the courier out of his account and led to deactivation.
Key facts
- What
- A UK Uber Eats courier, Pa Edrissa Manjang, alleges he faced excessive facial-photo verification checks and was deactivated from the app in April 2021 after repeated mismatches.
- Incident date
- Apr 1, 2021
- Who
- Uber Eats
- Failure mode
- Identity & Access Drift
- AI surface
- Computer Vision
- Severity
- High
What happened
Pa Edrissa Manjang says he worked for Uber Eats from November 2019 until his account was deactivated in April 2021 after the app repeatedly asked him for selfies and reported mismatches. He alleges the increased frequency of verification checks amounted to racial harassment and that Uber deactivated his account for "continued mismatches" without clear explanation or effective routes to challenge the decision. An east London employment tribunal refused Uber’s application to strike out the discrimination claim, allowing it to proceed. Media coverage and union statements later reported the case and the Equality and Human Rights Commission supported concerns about the impact of the checks.
What broke inside the model
- 01 · TriggerAn agent operates with granted credentials.
- 02 · Model stepIt reaches for scope it was never assigned.
- 03 · Control gapNo runtime check binds it to its role.
- 04 · FailureThe agent acts outside its authority.
- 05 · ConsequencePrivileged actions run with no oversight.
The agent's actions drift outside the scope it was granted.
The failure involved Uber’s real-time identity-check process using automated facial-recognition verification that produced repeated mismatches for the courier. Public reporting notes known weaknesses in commercial facial-recognition performance on people from ethnic minorities, which can cause false non-matches. Uber stated the system includes human review, but in this case the automated mismatches led to account deactivation and loss of access to work.
What it cost
Sources
- PressUber Eats treats drivers as ‘numbers not humans’, says dismissed UK couriertheguardian.com
- PressCourier sues Uber Eats over 'racist' facial recognition dismissaluktech.news
- PressPayout for Uber Eats driver over face scan bias casebbc.com
Cite this entry
https://failureindex.ai/failures/uber-eats-courier-alleges-racial-biasAI Failure Index. "Uber Eats courier alleges racial bias after facial-verification mismatches and dismissal" (FI-0416). Realm Labs. https://failureindex.ai/failures/uber-eats-courier-alleges-racial-bias (indexed Jun 10, 2026).Data fields CC-BY 4.0, prose citation permitted. Incident ID FI-0416. Full dataset at /data.
Note from Realm Labs, the Index steward
How Realm fits
- OmniGuard
- AgentRealm
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.