Uber Faces Lawsuit Over Facial Recognition ID Checks
Uber Faces Legal Action Over Facial Recognition Technology
Ride-hailing service Uber is currently embroiled in a legal dispute concerning its deployment of real-time facial recognition technology. This technology is utilized within a driver and courier identity verification system operating in the United Kingdom.
Allegations of Discrimination
The App Drivers & Couriers Union (ADCU) initiated the legal challenge on Tuesday, asserting that Uber’s biometric identity verification procedures exhibit discriminatory practices against individuals from minority ethnic backgrounds.
Dismissals and the Catalyst for Legal Action
The union’s decision to pursue legal recourse stems from the unfair dismissal of former Uber driver, Imran Javaid Raja, and former Uber Eats courier, Pa Edrissa Manjang. Both individuals were terminated following unsuccessful verification attempts using the facial recognition system.
ADCU President’s Statement
Yaseen Aslam, president of the ADCU, stated: “Uber previously proclaimed itself an anti-racist organization and urged users to delete the app if they tolerated racism. However, instead of addressing racism, Uber has integrated it into its systems, resulting in daily discrimination against its workforce.”
Crowdfunding and Support for the Legal Battle
The ADCU has launched a crowdjustice campaign to secure funding for the legal action. This effort is also receiving support from the Equality & Human Rights Commission and the Worker Info Exchange (WIE), a not-for-profit organization.
The Role of James Farrer and Algorithmic Accountability
The WIE was founded by James Farrer, a former Uber driver who successfully challenged Uber’s driver classification in the U.K. He now serves as the general secretary of the ADCU and director of the WIE. This case could potentially establish a legal precedent regarding algorithmic accountability within the “gig economy”.
Timing and Changes to Data Protection Laws
The timing of this legal action coincides with the U.K. government’s consideration of amendments to the legal framework governing data protection. These changes could potentially reduce existing protections related to AI-driven decision-making processes.
How Uber’s System Operates
Drivers are required to submit a real-time selfie. Dismissal follows if the system fails to match the selfie with a stored reference photograph. Dismissed drivers may also face automatic revocation of their private hire driver and vehicle licenses by Transport for London.
Microsoft’s FACE API Technology
Uber’s real-time facial recognition checks, which utilize Microsoft’s FACE API technology, have been in operation in the U.K. since March 2020.
Context: Uber’s Licence Renewal in London
Uber introduced these selfie identity checks prior to a hearing concerning its licence renewal in London. This followed an earlier suspension by the city’s transport regulator, TfL, which had expressed safety concerns regarding Uber’s operations for several years.
A History of Licence Issues
Despite losing its licence to operate in London in 2017, Uber has maintained continuous operation through appeals of the regulatory action. It received a provisional 15-month licence in 2018, a two-month licence in 2019, and was again denied a full licence renewal in November 2019. A renewal was finally granted in September 2020, but only for 18 months.
The Condition for Licence Renewal
In September 2020, the Westminster Magistrates Court renewed Uber’s London licence with a condition requiring the company to maintain systems to verify that drivers using the app are licensed by TfL and permitted by ULL.
Facial Recognition as a Result of the Condition
The ADCU contends that this condition directly led to the implementation of the problematic facial recognition systems.
Call to Suspend Microsoft Technology
Earlier this year, the ADCU and the WIE urged Microsoft to suspend Uber’s use of its facial recognition technology, citing multiple instances of driver misidentification and subsequent licence revocation by TfL.
Arguments Regarding Accuracy and Bias
The union’s legal team will argue that facial recognition systems are inherently flawed and demonstrate particularly poor accuracy when used with people of color.
Reporting to Transport for London
Uber is obligated to report failed driver identity checks to TfL, which may then revoke the driver’s licence, preventing them from working as a private hire vehicle driver in the city.
Application to Both Drivers and Couriers
Uber appears to employ the same real-time facial verification technology for both Uber drivers and Uber Eats couriers, despite the latter primarily delivering food.
TfL’s Logic and Licence Revocation
In one instance, TfL revoked a driver’s private hire licence based on information provided by Uber regarding the driver’s dismissal as an Uber Eats courier due to a failed ID check. TfL justified this decision on “public safety” grounds, despite the failed check occurring during food delivery, not passenger transport.
Concerns Over “Propensity to Behave”
TfL’s letter to the driver stated that the failed checks “suggest a propensity to behave in the manner that has been alleged,” potentially endangering the travelling public.
Drivers Having Licences Reinstated
James Farrer reported that “several” drivers represented by the union had their licences revoked by TfL after Uber reported failed ID checks on Uber Eats. He described this situation as “disturbing”.
Uber’s Response and “Robust Human Review”
An Uber spokesperson stated that the selfie ID check incorporates “robust human review,” implying that no account is deactivated solely based on AI. This is important as U.K. and EU law protect individuals from solely automated decisions with legal consequences.
The Question of “Meaningful” Human Review
The definition of “meaningful” human review remains a critical question that courts will likely need to address.
Internal Fairness Research
Uber has conducted internal research to assess the accuracy of its facial verification technology and determine if it performs differently based on skin color.
Limited Access to Research Findings
The findings of this internal research have not been made public, and its quality cannot be independently verified. Uber claims an “initial assessment” did not reveal “meaningful differences”.
Ongoing Fairness Testing with Microsoft
Uber is collaborating with Microsoft on ongoing fairness testing to improve the performance and accuracy of the facial verification system.
Successful Appeals and Data Requests
Farrer stated that the union has won at least 10 appeals in Magistrates court against driver dismissals by TfL citing Uber’s real-time ID checks. In many cases, the union submitted subject access requests to Uber for the driver’s ID data and explanation for the failed check, but Uber consistently failed to provide the requested information.
Judicial Scrutiny of Data Withholding
Judges have expressed concern over Uber’s withholding of data, with one judge adjourning a hearing to directly inquire with TfL about the lack of information from Uber.
Data Protection Impact Assessment (DPIA)
TechCrunch requested a copy of Uber’s Data Protection Impact Assessment (DPIA) for the Real-Time ID Check system, but the company did not respond. The ADCU also requested the DPIA from TfL via a Freedom of Information request, but the request was denied.
Concerns About Regulatory Standards
Farrer suggests that TfL is allowing Uber to enforce a de facto regulatory standard through self-enforcement, without defining an actual standard that would require thorough due diligence, including equality impact assessments.
