Who Is Liable for Damage Caused by Humanoid Robots?

Who is liable for damage caused by humanoid robots in public? This question is becoming increasingly important as autonomous systems become more widespread. Humanoid robots are used in retail, at trade fairs or in train stations. At the same time, the risk of personal injury or property damage increases. Legally, the situation is complex because artificial intelligence can act independently. However, one principle still applies: robots are not legal entities. A person or a company is always liable. What matters is who caused the error or controlled the danger. Below you will find all relevant liability constellations under German and European law.

Key points at a glance

Humanoid robots are not independent liability subjects.
Manufacturers are liable for product defects according to product liability law.
Operators bear responsibility for traffic safety obligations.
Users are liable for improper or manipulated use.
New EU rules such as the AI Act tighten documentation and security requirements.

Who is liable for damage caused by humanoid robots in public?

Depending on the cause, the manufacturer, the operator, the software provider or the user are liable. The robot itself is not a legal entity. The basis is the Product Liability Act, Section 823 of the German Civil Code (BGB) and new EU regulations such as the AI Act.

Manufacturer liability and product liability law

If a humanoid robot causes damage due to a technical error, the manufacturer is usually liable. The basis is the Product Liability Act (ProdHaftG). The EU Product Liability Directive and Section 823 of the German Civil Code also apply. What matters is whether there is a product defect. Such an error can be in design, production or programming. Typical cases are software errors that lead to uncontrolled movements. Even faulty sensors cannot correctly detect people. If safety mechanisms fail, liability also arises. What is particularly important is that manufacturer liability is strict liability. This means that you do not have to prove your own fault. The only decisive factor is the defective condition of the product.

Operator liability and traffic safety obligation

If a company publicly uses a humanoid robot, it is responsible for its safe operation. The basis is the so-called traffic safety obligation. This is derived from Section 823 of the German Civil Code (BGB). Anyone who opens a source of danger must protect third parties. This applies, for example, in retail, at trade fairs or in train stations. Lack of maintenance can constitute a breach of duty. Inadequate security precautions are equally problematic. If a robot is used despite known risks, the liability risk increases significantly. Operators must carry out regular checks. They also have to secure danger areas. If damage still occurs, you are liable for breach of duty.

Operator liability by analogy with road traffic

In the legal discussion, owner-like liability for autonomous systems is being discussed. The model is Section 7 StVG from road traffic law. The vehicle owner is liable regardless of fault. There is currently no such special regulation for humanoid robots. Nevertheless, there are already special regulations for autonomous vehicles. As the autonomy of humanoid systems increases, the pressure for reform grows. Independent owner liability could be introduced in the future. The aim would be a clear risk allocation. However, general liability rules currently remain. Courts examine on a case-by-case basis who had actual control.

Liability of software providers and AI developers

A lot of damage is not caused by hardware, but by AI algorithms. This raises the question of the liability of software developers. In practice, the company behind the software is usually liable. The system integrator may also be responsible. The manufacturer may also be held liable if he uses the software. An individual developer is only liable in the event of gross negligence or intent. What matters is whether the system was implemented incorrectly. The state of the art also plays a role. If the error was foreseeable, the risk of liability increases. However, when it comes to complex AI systems, providing proof is often difficult.

User liability in the event of misconduct

It’s not just manufacturers or operators who are liable. The user can also be responsible. This is especially true if used improperly. If a security system is manipulated, it is your own fault. The same applies when used outside the intended environment. If the user ignores operating instructions, he bears the risk. Liability is then based on general principles of tort law. The decisive factor is whether the damage was caused by breach of duty. This is particularly relevant for private or commercial owners. Insurance can cover the risk, but does not replace liability testing.

New EU regulations and autonomous decisions

With the AI Act, the EU is creating a uniform legal framework for artificial intelligence for the first time. The AI Act distinguishes risk classes of AI systems. It regulates transparency obligations and security requirements. There are also extensive documentation requirements. Humanoid robots fall under this if they use AI. At the same time, the EU Product Liability Directive is being modernized. The aim is to ease questions about the burden of proof in cases of AI damage. Fully autonomous decisions are particularly problematic. Here courts examine the state of the art. Predictability also plays a central role. In addition, it is assessed whether reasonable security measures have been implemented.

Who is specifically liable? Overview

Cause of damage	Liability
Technical defect	Manufacturer
Lack of maintenance	Operator
Incorrect usage	Users
Software error	Manufacturer / Provider
Unclear autonomy decision	Legally still changing

Legal update 2026: Product liability also includes software

The new EU Product Liability Directive expressly treats software as a product. Manufacturers can therefore be liable not only for mechanical defects, but also for faulty software, security updates and inadequate cybersecurity. In the case of learning systems, it is also relevant how predictable behavior develops after it has been placed on the market.

This does not replace operator responsibility. Anyone who uses a robot must use it as intended, maintain it, assess risks and limit risks to third parties. In a specific case of damage, liability depends on the cause, contractual relationships and evidence; There is currently no blanket “robot owner liability” as with motor vehicles.

Comparative liability: UK, US and India

European product-liability rules do not automatically apply worldwide. UK claims may combine the Consumer Protection Act 1987, negligence, contract and workplace duties. US product-liability law is primarily state-based, so defect tests and available damages differ. India’s Consumer Protection Act 2019 expressly provides product-liability actions against manufacturers, service providers and sellers. Across all three systems, logs, maintenance records, warnings, updates and evidence of foreseeable misuse can determine who bears responsibility.

Conclusion

Who is liable for damage caused by humanoid robots in public? The answer depends on the cause. Manufacturers, operators, software providers or users come into question. The robot itself remains legally meaningless. However, with increasing autonomy, the demands on companies increase. The AI Act and product liability reform increase responsibility. One thing is clear: Liability is increasingly shifting to those economically responsible – not to the machine.

Bewerte den Beitrag hier!

[Total: 0 Average: 0]

Nico Nuss

Website

Author Nico Nuss has been working on mobile computing and automation software since 2001. Drawing on his experience and strong interest in future technologies, he focuses on robotics and AI.