I had a discussion about drones for civilian use, e.g. self-driving cars without a human at its wheels, transport devices that would fly products you otherwise would buy in the supermarket directly to your home, and robots that collect trash from the streets all by themselves.

The most controversial issue was of who takes the responsibility for damages caused by such a drone. A list of potential candidates were:

  • since the drones acts on its own, the drone itself
  • the developer of that malfunctioning code
  • the company selling the drone
  • the owner of the drone

I find myself not agreeing with any of those possibilities by themselves, yet agreeing on some on some levels.

I think there is no such thing as malevolent intent for a drone, so every damage would be an accident. I hence doubt we could treat such a drone like a human and have it put on trial.

An accident then could either be caused by a force majeure, e.g. due to a natural disaster, or be boiled down to technical failure.

For technical failure, it would be either the the company's fault for not testing the device enough or the developer's one for making a mistake. The severity would then be determined if it was a somewhat understandable oversight or a grossly negligent act.

I think the real question hiding in there is: What kind of security standard do we want to apply as a society on such devices without hindering innovation too much? Or don't we want them at all?

If a company or a developer must fear high compensation for damages, wouldn't the rational choice be not to build such drones?

  • Are you deliberately not considering the use of drones in war, defence & surveillance - as in the border areas between Afghanistan and Pakistan? That is you're only considering the use of drones within the domestic & civilian sphere? Commented Jun 11, 2013 at 6:24
  • 1
    Here's some related stuff: plato.stanford.edu/entries/technology/#DevEthTec. And, without me having any further thoughts about it, you might draw an analogy with children damaging stuff. :)
    – user3164
    Commented Jun 11, 2013 at 8:37
  • I must have skipped over your first sentence 'I had a discussion about drones for civilian use' - so you are talking about the domestic sphere :). Commented Jun 11, 2013 at 11:34
  • also I think the drones used in war are operated by a human, not automatical
    – mart
    Commented Jun 18, 2013 at 9:18

3 Answers 3


In my eyes the answer is simpler than one might think:

  • If the damage is caused by a failure of the hardware or the software shipped with the robot, the company selling the drone has to be held responsible (within the warranty). Whether the company wants to sue the developer depends is up to the company and depends on the contract that the developer signed. This case corresponds to typical warranty case where a car that was produced with defects causes a damage.

  • If the damage is caused because some parts have worn out and the owner did not take care of this (after the warranty), the owner can be held responsible. This corresponds to a car owner not taking care of his tires which leads to an accident.

For the next point it is important do distinguish different cases of controller:

  • Remote control: If the robot is a drone and remote controlled by a human, the operator is responsible for any damages. Corresponds to guy driving a car.

  • Autonomous static control: If the robot is executing a fix software which did not get affected by any external causes, then the company can be held responsible for damages. Corresponds to a broken product.

  • Autonomous supervised learner: If the robot is capable of learning, then the user teaching the robot is reliable if it can be shown that other robots which were trained in a better way don't cause any damages. This corresponds to not worn out tires just on the controller side. If the robot is doomed to fail i.e. it causes a damage independent of the training - then the company is responsible for the damage. Corresponds to a broken product

  • Autonomous unsupervised learner: If the robot is doomed to fail, it's the companies responsibility. If a "bad influence" of the owner on the behavior of the robot can be proven, then the owner is responsible. If this is not the case, the damage was caused by a unforeseen constellation in the environment of the robot so that it learned a bad behavior so nobody is responsible (robot has to be reset or destroyed). This is similar to a natural disaster.

  • 2
    Although useful, this does not answer the question, but resembles a mere legal opinion. That is, it only addresses current law, whereas the question seems to focus on what future law should look like.
    – user3164
    Commented Jun 20, 2013 at 17:49
  • Point 1 - why is what you call a legal opinion not an answer for a question with a juristic touch? Point 2 - why should future law not look like the current one? Btw. I am not a lawyer but a researcher in the field of neuroinformatics and the core of my answer is that we don't need a new quality of responsibility because robots either behave like tools/machines or like kids/animals and they are already part of out ethics/laws.
    – Brandli
    Commented Jun 20, 2013 at 18:31
  • I said "resembles", not "is". If your point is that current law is perfectly fine with respect to "drone innovations", then that's fine with me (although you might include that explicitly in your answer; perhaps you might also comment on when tools-law applies and when kids-law applies). PS: The phrasing "our ethics" seems somewhat contentious.
    – user3164
    Commented Jun 20, 2013 at 18:47
  • I would say that a legal opinion is exactly what is sought - the question is not "Who should we hold responsible," but "Who IS responsible" - which is a question for the court.
    – shieldfoss
    Commented Jun 23, 2013 at 22:26

I agree with all of Brandli's points, but I would just like to add another subject of blame or cause: society. As well as manufacturers of a dangerous machine, and irresponsible users of a potentially dangerous machine, we can also point the finger at the government or culture that let either of those situations occur.

Examples of how this is handled already:

  • The transport authority will not award driving licenses to people whose sight is so poor as to make them a danger to others.

  • There are probably restrictions on gun ownership for people with mental health problems.

However regulation is a restriction on freedom, so it should be done very carefully. The idea of compensation-punishment seems like a more liberal approach. Although the culture of suing for damages has its disadvantages too, and I'm sure there is a debate to be had there.


Great question ! In order to place blame at someone's feet, the legal case would need to show they were negligent.

I guess the example of a self-driving car smacking into something, the legal case for the 'victim' would be that the company programming the car should have considered such situations and were negligent in allowing their car on the road without such consideration. I'm not saying that would hold up- there would be test cases etc as it's a new situation for the world, but that would probably be the argument.

Bottom line is that if a company releases a device into the world, they can't absolve themselves of the responsibility if things go wrong. Only in the case where it's used wrongly, or not as intended by hte manufacturer, would the responsibility lie with the person operating the device.

So in each case I guess the legal team(s) would establish why the incident occurred and who it was that did something that led to that behaviour. ., either the drone manufacturer, the company/person operating/programming it, or maybe someone messing with it while it's out and about.

Bottom line: Yes, companies releasing devices into the world have to face up to their responsibilities, and pay the price if they don't show due dilligence. Nothing new there.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .