Artificial Intelligence and its Tax Consequences on Worker Classification

Blog by Gianluca Ritz, Junior Associate

If you were told to think of AI as it is right now, you would immediately think of the new wave of generative AI algorithms like ChaptGPT, DALL-E, and Gemini. While these are the new exiting tools, AI has been around for a long time in one form or another. There is not a common definition, but generally, AI is a series of equations, given a set of human-defined objectives, and which uses human or machine based inputs to perceive its environment to create an output which comports with the objective.[1] AI is all around you, it drives what content you see on YouTube and what shows up on your TikTok feed, it is also is responsible for almost every function of  popular app-based services like UberEats and DoorDash.[2] This includes payment, job offers to drivers, preferred routes, refunds, and worker discipline.[3]

While its prevalence is clear, the weight of AI in the eyes of the law – especially worker classification – is still not. This is especially pertinent given the recent large increase in platform/contractor work, which has much to do with these app-based services. [4]  Their lower barrier of entry and pick-up and go nature means that many offer services through these apps to supplement their more stable income. [5] Why people are able to work so quick on the app has to do with the tax scheme.  In most parts of the world, and especially the United States, working for these apps makes you an independent contractor,[6] which means companies like Uber do not have to worry about paying many of the costs that comes with an employee. App-based services do not have to pay towards minimum wage, social security taxes, overtime pay, and unemployment insurance. In exchange for the lack of protection, you can pick up and play; all you often need is a car and to pass a criminal background check.  As well, Uber Drivers are often able to choose their hours and location of work. But, again, this is all predicated on app-based workers not receiving essential labor protections.

A large reason why this tax scheme has been so successful has been because many courts have refused to see these workers as employees, and a lot of that has to do with Artificial Intelligence. Many labor laws in many countries are archaic, and many legal tests for employee status presume a human factor or director.[7] AI confuses matters greatly, because most of its inputs come not from the employer, but customers. Worse, these are proprietary technologies, so the exact equation, inputs tracked, and their relative weight is not told to anyone unless the company responsible reveals it.[8] This makes algorithmic management an invisible force, and often hides the level of true control these companies have over their workers.  This secrecy has seemed to frustrate EU efforts to make companies more responsible. There is already protective legislation in place that allows app-based workers to sue for decisions made by an algorithm that significantly affect them, however, it has proven difficult to prove due to a lack of data reporting requirements.[9]

The result of all this has been confusion on the part of courts as to how AI should be weighed on the classification scale. Many only look to surface level factors of control, such as if the driver can control their hours of if they can refuse jobs.[10]

However, despite these challenges, many courts have taken the initiative, and named workers for apps like Uber and Lyft as employees. For instance, California passed Assembly Bill 5, which made all workers employees (although it was later amended to specifically exempt app-based drivers).[11] The idea was that unless you could prove you were truly independent of the platform (i.e. you were using the services of the platform to just market your own services) then you were a worker. France and Spain also both passed laws stating that Uber drivers are employees.[12]

The Netherlands did likewise, and a quote from the case was rather interesting:

“In today’s technology-dominated age, the criterion of ‘authority’ has taken on a more indirect (often digital) monitoring function that deviates from the classical model. Employees have become more independent, and conduct their work at more variable (self-chosen) times . . . The algorithm of the Uber app then determines how the rides are allocated and prioritized. It does this on the basis of the priorities set by Uber . . . . it was stated on behalf of Uber that Uber – to put it simply – can ‘control the buttons of the app’ and change the settings. “[13]

It is a commonsense approach; in that it is rather silly to separate the decisions of an algorithm and the company. An independent contractor should be independent of the platform, and an algorithm that offers any kind of management defeats the purpose. The United Kingdom reached a similar conclusion, offering that if the app works to create a standardized product, the app is not an information society service, and their workers are employees. [14]

With that said, the dominant tax scheme for these app-based services is still based on their workers being independent contractors – which limits their labor protections. However, if the recent cases and the rise of AI legislation are any indication, this will most likely change. As data becomes more open, and judicial literacy of these tools increase, it will likely be the case that these app-based services will be singled out, and these drivers will finally get the protections of any other employee.

[1] Marguerita Lane and Anne Saint-Martin, The impact of Artificial Intelligence on the labor market, No. 256 OECD Social, Employment and Migration Working Papers 1, 17 (2021).

[2] See e.g., Paul Covington et. al. Deep Neural Networks for YouTube Recommendations, Sept 2016 RecSys ’16: Proceedings of the 10th ACM Conference on Recommender Systems, 191 (2016).

[3] See Fedratie Nederlandse Vakbeweging v. Uber B.V., 8937120 CV EXPL 20-22882, Judgement of the subdistrict courts on whether Uber Drivers are covered by the Collective Labor Agreement for Taxi Transport,  ¶  3-9 (Rechtbank Amsterdam, Sept. 13, 2021) (hereinafter FNV v. Uber).

[4] Freelance, side hustles, and gigs: Many more Americans have become independent workers, MᴄKɪɴsᴇʏ & Cᴏᴍᴘᴀɴʏ (Aug. 23, 2022),

[5] Id.

[6] Megan Carullo, Uber drivers are independent contractors, not employees, says National Labor Relations Board,  CBS Nᴇᴡs (May 14, 2019),

[7] John A. II Pearce & Jonathan P. Silva, The Future of Independent Contractors and Their Status as Non-Employees: Moving on from a Common Law Standard, 14 Hastings Bus. L.J. 1, 2 (2018).

[8] Proposal for a Directive of the European Parliament and of the Council on improving working conditions in platform work, COM (2021) 762 final (Apr. 14, 2021) (hereinafter EU PWD).

[9] Id.

[10] Daniel Wiessner, U.S. Judge says Uber drivers are not company’s employees, Rᴇᴜᴛᴇʀs (Apr. 12, 2018),

[11] AB5 Statute, CA Labor and Workforce Development Agency (Sept. 19, 2019),

[12] See Marten Männis, Uber Drivers considered as employees in France, court rules, INHOUSE LEGAL (Mar. 6, 2020),; see also Leonie Carter, Spain approved a law protecting delivery workers. Here’s what you need to know, POLITICO (May 11, 2021), (reporting that the Spanish cabinet passed a law that requires online delivery platforms operating in the country to classify their couriers as employees rather than independent contractors).

[13] FNV v. Uber, supra note 3 at ¶ 26-28.

[14] Uber BV v. Aslam [2021] UKSC 5 at ¶ 99-104

Leave a Reply

Your email address will not be published. Required fields are marked *