If your platform uses software to suspend accounts, flag fraud, rank users or cut off access to a service, you’ll need to pay attention to the €825 million fine imposed on Uber in August 2026. It is the second-largest penalty ever issued under the GDPR, and it was not triggered by a data breach, a cookie banner or an unlawful data transfer. It was triggered by an algorithm making decisions about people's livelihoods without a human being looking at them first.
For technology businesses operating in Europe, the decision is a clear statement of how regulators intend to apply existing data protection law to automated systems, well before the heavier obligations of the EU AI Act arrive. This article explains what Uber was fined for, why the amount is so large, what the case tells us about the direction of enforcement, and what you should be reviewing in your own products.
What Did the Regulators Decide in the Uber Case?
On 21st August 2026, the Dutch Data Protection Authority, the Autoriteit Persoonsgegevens (AP), fined Uber B.V. and Uber Technologies Inc. €824,990,000 for unlawful automated decision-making. The AP led the case because Uber's main European establishment is in Amsterdam, but it worked closely with France's data protection authority, the CNIL, which published its own announcement of the fine because the case began in France.
The investigation originated in a collective complaint filed with the CNIL in 2020 by the Ligue des droits de l'Homme on behalf of more than 170 French drivers, later supplemented in 2021. The regulators found that, between 2018 and 2022, Uber's systems deactivated drivers' accounts without human intervention in two situations: temporary suspensions where the system suspected fraud, and temporary or permanent deactivations where drivers' customer ratings fell below a threshold. In each case the consequence was immediate, because a deactivated driver could not accept rides and therefore could not earn.
The AP concluded that these were decisions based solely on automated processing that significantly affected the drivers, contrary to Article 22 of the GDPR. Its deputy chair put the principle in simple terms: a computer should not make decisions with major consequences for someone on its own, and those decisions should have been looked at by a person first.
How Has Uber Responded?
Uber has reportedly said it will appeal. According to published reports, it disputes that the decisions were fully automated, argues that fraud-related suspensions were brief, says that only a small number of drivers were permanently deactivated in Europe in 2021, and says it has since introduced human review and appeal processes. Uber is also still contesting two earlier penalties: a €10 million fine from the AP in December 2023 for failing to give drivers adequate information, and a €290 million fine in July 2024 for transferring European drivers' data to the United States without appropriate safeguards. The decision may therefore be varied on appeal, and it should be read as the regulator's position rather than a final judicial ruling.
Why Was the Dutch Data Protection Authority Fine So Large?
GDPR fines for the most serious infringements, including breaches of the rights of data subjects under Article 22, can reach 4% of a group's total worldwide annual turnover. For a company of Uber's size, that ceiling runs to well over a billion euros, and reported analysis suggests the penalty represents a little under 2% of annual turnover. In setting the amount, regulators consider the nature and duration of the infringement, the number of people affected, the seriousness of the consequences for them and the company's history.
The conduct lasted around four years and it affected people's ability to earn a living, and the company had already been sanctioned twice in connection with its treatment of drivers' data. The size of the fine also sends a deliberate signal to the wider market that automated decisions about people's access to work, income or essential services are now an enforcement priority in their own right.
What Does “Human Intervention” Actually Require?
The most useful lesson from the case concerns the difference between having a human somewhere in the process and having meaningful human involvement in the decision. Article 22 prohibits decisions based solely on automated processing that produce legal effects or similarly significant effects, subject to limited exceptions such as contractual necessity, explicit consent or authorisation by law. Even where an exception applies, the business must provide safeguards, including the right to obtain human intervention, to express a point of view and to contest the decision.
The European Data Protection Board's guidance has long made clear that a human rubber-stamping an algorithm's output does not take a decision outside Article 22. The reviewer needs real authority to change the outcome, enough information and time to assess it, and the competence to do so. The Court of Justice has reinforced this direction in the SCHUFA judgment in 2023, which held that an automated score can itself be a decision where a third party relies heavily on it, and in Dun & Bradstreet Austria in 2025, which confirmed that individuals are entitled to an explanation of the procedure and principles actually applied to reach an automated decision.
Taken together, the Uber decision and these judgments suggest that regulators will test what happens in practice rather than what a policy says. A business that tells regulators a human reviews every suspension, but whose logs show that suspensions take effect before anyone looks at them, is in a weak position.
Is This Only a Problem for Gig Economy Platforms?
No, although platforms that manage workers through apps are the most obvious candidates. The same analysis applies to any product that uses automated logic to make significant decisions about individuals. Examples include a SaaS marketplace that suspends sellers automatically when a fraud score is triggered, a fintech that declines credit or freezes accounts based on a model, a subscription service that permanently bans users on the basis of automated abuse detection, a recruitment tool that rejects candidates without a recruiter seeing the application, and an HR platform that flags employees for disciplinary action based on productivity metrics.
It is also important to remember who is responsible. Where a US or UK software provider supplies an automated decision tool to a European customer, the customer that uses the output to make decisions will often be the controller with primary responsibility under Article 22. That does not insulate the provider commercially. Enterprise customers are increasingly asking vendors to demonstrate that their products support genuine human review, provide explanations and keep audit trails, and contracts are allocating that risk accordingly.
How Does This Fit with Wider EU Regulation of Algorithmic Management?
The Uber decision sits alongside two further EU regimes that are raising the bar on automated decisions about workers. The Platform Work Directive, which Member States must implement by 2nd December 2026, introduces specific rules on algorithmic management for digital labour platforms. It requires platforms to inform workers about automated monitoring and decision-making systems, to ensure human oversight of those systems, and to have decisions to restrict, suspend or terminate an account taken by a human being. In other words, the conduct at the centre of the Uber case is about to be addressed expressly in employment-related legislation across the EU, with France and other Member States each implementing the rules in their national law.
The EU AI Act classifies AI systems used for recruitment, task allocation, performance monitoring and decisions affecting the termination of work relationships as high-risk. Following the Digital Omnibus on AI, which entered into force in July 2026, those high-risk obligations now apply from 2 December 2027 rather than August 2026. The delay has led some businesses to deprioritise algorithmic governance. The Uber fine demonstrates why that would be a mistake: the GDPR already applies, it is already being enforced, and the penalties are already substantial.
What Does This Tell Us about the Direction of Enforcement?
Three trends stand out. First, European regulators are using the GDPR as the primary tool for policing automated systems while AI-specific legislation is phased in, which means that businesses cannot treat AI governance as a future project. Second, collective complaints brought by civil society organisations on behalf of groups of affected individuals are producing large penalties, and the Representative Actions Directive, together with national collective redress mechanisms, is making this route more accessible. Third, enforcement takes years but it arrives. The Uber complaint was filed in 2020 about conduct dating back to 2018, and the fine was imposed in 2026. Decisions your systems make today may be examined long after the product has changed.
There is also a French dimension worth noting for businesses expanding into France. The CNIL has been particularly active on questions of worker surveillance and algorithmic management, and French trade unions and rights organisations have shown a willingness to bring complaints. A business that manages French workers or contractors through automated tools should expect those tools to receive scrutiny.
What Should You Review Now?
Start by identifying every point in your product or operations where software makes, or effectively makes, a decision that significantly affects an individual, such as suspending an account, refusing a service, setting a price for an individual, ranking a candidate or flagging someone for action. For each one, establish whether a person genuinely reviews the decision before it takes effect, whether that person has the authority and information to change it, and whether you can evidence that review from your logs.
Then look at what you tell people. Individuals affected by automated decisions must be told that the processing exists, given meaningful information about the logic involved and its likely consequences, and offered a way to contest the outcome. Your privacy notices, user terms and account suspension messages should be consistent with how the system actually works.
If you supply automated decision tools to European customers, review your product documentation and contracts so that customers can meet their own obligations, and consider whether your default settings support human review rather than bypassing it. Where your product operates as or supports a digital labour platform, the Platform Work Directive deadline of 2nd December 2026 makes this review urgent.
The right answer will depend on how your system works, what decisions it influences and where your users are based. Regulators have now shown how they will approach these questions, and it is considerably easier to build meaningful human review into a product at the design stage than to retrofit it under investigation.
How Can Gerrish Legal Help?
Gerrish Legal is a dynamic digital law firm. We pride ourselves on giving high-quality and expert legal advice to our valued clients. We specialise in many aspects of digital law such as GDPR, data privacy, digital and technology law, commercial law, and intellectual property.
We give companies the support they need to successfully and confidently run their businesses whilst complying with legal regulations without the burdens of keeping up with ever-changing digital requirements.
We are here to help you, get in contact with us today for more information.