The use of AI is becoming increasingly prevalent in the workplace. For example, AI can assist during the recruitment and hiring process by scanning CVs and application forms to prioritise applications using certain keywords. It may also assist as part of performance management with AI tools being available to allocate tasks and schedule shifts. The outsourcing of tasks such as these, which were previously carried out by humans, gives rise to potential new legal issues within the employment law context. The case of Manjang v Uber Eats Ltd and others ET/3206212/21 is one of the first employment claims involving AI and automated decision-making.

Mr Manjang, who is a black male of African descent, commenced work as a driver for Uber’s Ubereats food delivery service in November 2019. In April 2020, Uber Eats introduced Microsoft facial recognition software, which requires drivers to take a real time photograph of themselves (a ‘selfie’) for verification when using the app. Access to the app is a pre-requisite to accessing work and consequently getting paid. The purpose of the verification is to ensure that driver accounts are not shared, and if the photograph does not match the driver’s account profile picture, they risk having their account terminated.

In April 2021, Mr Manjang was removed from the platform following a failed facial recognition check and subsequent automated process. Mr Manjang was told by Uber Eats that they had found ‘continued mismatches’ in the photos of his face he had taken for the purpose of accessing the app.

In October 2021, Mr Manjang brought a claim for indirect race discrimination under section 19 Equality Act 2021, amongst others. Under section 19, A discriminates against B where:

1. i) A applies to B a provision, criterion or practice (PCP);
2. ii) B has a protected characteristic;

iii) A also applies (or would apply) that PCP to persons who do not share B’s protected characteristic;

1. iv) The PCP puts or would put persons with whom B shares the protected characteristic at a particular disadvantage compared to others;
2. v) The PCP puts or would put B to that disadvantage;
3. vi) A cannot show the PCP to be a proportionate means of achieving a legitimate aim.

Mr Manjang relied on a number of PCPs, one being the need to undergo facial recognition checks in general and/or by way of the software adopted by Uber Eats. It was asserted that the PCP put black/non-white people and/or people of African descent or origin at a particular disadvantage in that they are less likely to pass the facial recognition test and thus more likely to face the consequences of being barred from the app and/or employment with Uber.

There is, as yet, no caselaw as to whether machine learning algorithms underlying facial recognition software can be considered a PCP for the purposes of section 19(1) of the 2021 Act. However, a report for the Trades Union Congress by the AI Law Consultancy on the legal implications of AI systems in the workplace, concludes that it is likely to be considered a PCP.

Uber Eats applied to have the claim struck out but following a preliminary hearing in May 2022, Employment Judge Frazer dismissed the application. A final hearing was scheduled for 17 days in November 2024, but the parties have now reached financial settlement.

The full judgment of Employment Judge Frazer can be accessed here.

The report by the AI Law Consultancy can be accessed here.

For any other enquiries, or to discuss instructing the Civil Team at Atlantic Chambers, please contact clerks@atlanticchambers.co.uk