The scene is easy to imagine. A customer walks into a chic clothing store in Antwerp or a bustling electronics shop in Hamburg. They are greeted not by a person, but by a sleek, bipedal humanoid robot employee. “Welcome,” it says in a pleasant, synthesized voice. “I am the store’s robotic assistant. How may I help you today?”
This scenario, once the stuff of science fiction, is now teetering on the edge of reality. Companies like Unitree Robotics are producing increasingly sophisticated humanoid models. And businesses, squeezed by labour shortages and rising costs, are watching closely. But as we stand today, a critical question looms for any European entrepreneur dreaming of an automated workforce: Is it actually legal?
The answer is a resounding, and profoundly European, “Yes, but…”
Replacing a human salesperson with an autonomous robot is not a simple transaction. It’s a journey into a dense and newly-formed forest of regulation, where trailblazing laws on artificial intelligence intersect with long-standing rules on labour, liability, and data privacy. For any business owner contemplating this move, it’s less about a simple purchase. And more about launching a full-scale project in legal and operational compliance.
The AI Act: The New Rulebook on the Block
The single most important piece of this puzzle is the EU AI Act. As of today, its key provisions are now coming into force, establishing the world’s first comprehensive legal framework for AI. The Act doesn’t ban workplace automation. Instead, it creates a pyramid of risk, and your legal obligations depend entirely on where your robot employee sits on that pyramid.
For a humanoid salesperson, two scenarios are likely:
1. The ‘Limited Risk’ Assistant:
In this most probable scenario, the robot’s tasks are informational. It welcomes customers, checks stock levels, provides product details, and guides people through the store. Here, the AI Act’s primary demand is transparency. The law is crystal clear: a human must always know they are interacting with a machine. There can be no deception. This means your robot must be programmed to introduce itself as such, or wear clear visual identifiers. The goal is to prevent manipulation and preserve human autonomy.
2. The ‘High-Risk’ Decision-Maker:
The game changes dramatically if the robot’s role becomes more critical. If, for instance, it uses facial recognition to identify loyal customers (processing biometric data), autonomously decides on a customer’s eligibility for in-store credit, or performs active security surveillance, it is immediately classified as a “high-risk” AI system.
This classification is a regulatory tripwire. Deploying a high-risk system requires a mountain of compliance: rigorous conformity assessments, registration in an EU database, detailed technical documentation, robust cybersecurity, and, crucially, the guarantee of meaningful human oversight. An operator must be able to intervene or shut the system down at any time. For a small or medium-sized business, the cost and complexity of deploying a high-risk system could be prohibitive.
The Human Factor: Labour Law in the Age of Automation
Before you even get to the AI Act, there is the human employee the robot is intended to replace. European labour law, particularly in countries like Belgium and France, is robust in its protection of workers.
There is no law forbidding a company from automating a position. However, if this automation leads to a dismissal, that dismissal must follow strict legal procedures. This includes respecting legal notice periods, calculating and paying severance, and providing a valid justification for the termination—in this case, economic and organizational restructuring.
Furthermore, a significant change in work organization, such as introducing autonomous robots, requires social consultation. In Belgium, for example, an employer is obligated to inform and consult with their Works Council (Conseil d’Entreprise) or, in its absence, the Committee for Prevention and Protection at Work (CPPT). Simply showing up one Monday morning with a robot and a termination letter for an employee is a recipe for a legal dispute.
The Physical Reality: Safety, Liability, and the Rogue Robot Employee
A 100kg humanoid robot is not a software update; it’s a piece of heavy machinery operating in a dynamic, unpredictable public space. What happens when it bumps into a child, rolls over a customer’s foot, or knocks over a display of expensive crystal?
First, the robot itself must bear the CE marking under the EU’s Machinery Directive, a declaration from the manufacturer that it meets essential health and safety requirements. But the responsibility doesn’t end there.
The new AI Liability Directive, designed to work in tandem with the AI Act, aims to make it easier for victims to get compensation when harmed by AI. It introduces a “presumption of causality,” meaning if an operator (the store owner) fails to comply with a rule (like ensuring proper oversight) and damage occurs, the court can presume that this failure caused the damage.
This leads to the most practical of considerations: insurance. A standard public liability policy is almost certainly inadequate. A business owner must have a frank conversation with their insurer to secure a specific policy that covers damages caused by an autonomous mobile robot. The cost of this specialized insurance will be a significant and recurring line item in any realistic budget.
The Data Dilemma: Your Robot is a Rolling GDPR Sensor
To navigate a store, your humanoid robot uses a suite of sensors: cameras, microphones, LiDAR. In doing so, it is constantly collecting data, including the images and voices of your customers—data that is strictly protected by the General Data Protection Regulation (GDPR).
This triggers a cascade of obligations. The business must have a clear legal basis for processing this data. It must inform customers via clear signage that they are being recorded. Crucially, for any new and potentially intrusive technology like this, a Data Protection Impact Assessment (DPIA) is almost certainly mandatory. This is a formal process where the business must analyze the risks to individuals’ privacy and detail the measures taken to mitigate them. It’s a complex document that often requires specialist legal advice to complete correctly.
The Verdict in 2025: Possible, But is it Practical?
So, can you replace your salesperson with a robot? Yes, legally, the path exists. But it is a path paved with compliance.
A business owner must navigate the AI Act’s risk tiers, the complexities of labour law, the new realities of AI liability, the stringent demands of GDPR, and the practical necessity of securing specialized insurance.
For now, the most likely and legally simplest application will be the “limited risk” robotic assistant, a novelty that offers basic help and information with full transparency. The vision of a fully autonomous retail workforce, capable of complex interaction and decision-making, remains on the other side of a formidable wall of regulatory and financial hurdles.
The European approach is clear: innovation is welcome, but it will not come at the expense of safety, fundamental rights, or transparency. Before you hire your first robot, your most important new hire should probably be a very good lawyer.