Loss of control over personal data, excessive surveillance, unjust automated decision-making, hidden metrics and untransparent algorithmic management: all forms of workplace data abuse that could become considerably harder to prove. These were the findings of a study examining the proposed Digital Omnibus done by the European Trade Union Institute.
The Digital Omnibus is currently being discussed in the EU parliament. It is a package proposed by the European Commission targeting among other laws the General Data Protection Regulation (GDPR), which is the main EU personal-data law. For example, the proposal could restrict GDPR data access requests for purposes beyond data protection, such as gathering evidence for an employment dispute.
The Commission presents the new framework as a simplification and competitiveness measure. It estimates that the package could deliver up to €5 billion in administrative cost savings by 2029. More than 100 European civil society organizations have called the proposal „an attempt to covertly dismantle Europe’s strongest protections against digital threats“.
What could those changes mean for workers trying to challenge the way their data is used? As the EU commission did not carry out an impact assessment for the Digital Omnibus, researchers Aida Ponce Del Castillo and Thomas le Bonniec conducted this study on behalf of the European Trade Union Institute.
To test the amendments’ possible effects, the authors applied them counterfactually to four workplace cases that had previously led to regulatory enforcement or successful legal claims against workplace data abuse. Cases involving the Taxi aggregator platforms Uber and Ola were among them.
Burden of proof shifts to employees
In one of the cases, six Uber drivers sought access to personal data and information about how the platform’s systems affected their work, including information on matching, pricing and ratings. Their aim was to understand how Uber’s algorithmic management systems were making or influencing decisions about them. A similar case was filed against Ola, a ridesharing company based in India.
In 2023, the Amsterdam Court of Appeal ultimately upheld claims by the drivers and required Uber and Ola to disclose specified personal data and meaningful information about certain automated decision-making systems.
The drivers used their existing GDPR access rights to obtain information that became important to their claims. The study’s authors argue that reaching similar outcomes could become more difficult under the Digital Omnibus changes.
Among other things, the amendments would create broader grounds for employers or platforms to resist workers’ requests for access to their personal data. An employer could, for example, argue that a worker was using a GDPR access request for a purpose other than exercising data-protection rights, such as gathering evidence for an employment dispute.
That distinction matters because employers and platforms often control the very data, algorithms and records workers need to understand what has happened to them. Workers often do not have access to the evidence needed to support their claims.
According to the study, the changes could therefore shift a greater burden onto workers. They may need to formulate more specific requests, challenge refusals or seek legal assistance simply to obtain information already held by their employer or platform.
Commenting on the proposed changes, study author Aida Ponce Del Castillo of the ETUI in Brussels said: “A worker may be right about unlawful surveillance and still be unable to prove it. If the EU makes it harder to obtain information from an employer, it also makes it harder for people to challenge what happens to them at work.”
Automated decisions harder to challenge
In a related case, four Uber drivers challenged the deactivation of their accounts after Uber systems flagged suspected fraudulent activity. The drivers argued that these were consequential decisions made solely through automated processing, without meaningful human intervention.
For three of the four drivers in the actual case, the court found that Uber had not demonstrated meaningful human intervention. The drivers had not been heard, and Uber had not shown that its reviewers had sufficient information and authority to change the automated outcome.
Alles netzpolitisch Relevante
Drei Mal pro Woche als Newsletter in deiner Inbox.
The GDPR provides protections against certain decisions based solely on automated processing. Currently, the availability of an equally effective human alternative can be relevant when assessing whether automated processing was really necessary.
The proposed amendments would change this test. The existence of a possible human alternative would no longer, by itself, be enough to argue that automated processing was unnecessary.
Applying that change to the Uber case, Uber would be in a stronger legal position, the authors argue. The company could argue that automated fraud detection was necessary for performing its contractual relationship with drivers, even if a human could have made the decision instead.
The authors conclude that under the proposed changes the legal pathway to comparable outcomes to the Uber / Ola cases would have been narrower: the drivers could have faced greater obstacles obtaining information, while Uber would have been in a stronger position to justify automated decision-making.
On top of an AI high-risks systems delay
On a similar note, a part of the Digital Omnibus package concerning AI was already adopted in July this year. It delayed the application of the AI Act’s requirements for high-risk systems by 16 months, until December 2027, including requirements affecting certain AI systems used in employment and worker management.
Taken together, the authors argue, the two sets of changes could reinforce each other: AI Act safeguards intended to prevent harmful practices would take effect later, while changes to the GDPR could make it more difficult for workers to investigate or challenge those practices after they occur.
In other words, workers could face fewer safeguards before an AI system affects them and greater hurdles in finding out what happened afterwards.

Schreibe eine Ergänzung!