Who's Responsible For Accidents Caused By Open Source Self-Driving Car Software? (ieee.org) 114
Here's the problem. "You could download Comma.ai's new open-source Python code from Github, grab the necessary hardware, and follow the company's instructions to add semi-autonomous capabilities to specific Acura and Honda model cars (with more vehicles to follow)," writes IEEE Spectrum. But then who's legally responsible if there's an accident?
Long-time Slashdot reader Registered Coward v2 writes:
While many legal experts agree OSS is "buyer beware" and that Comma.ai and its CEO Georg Hotz would not be liable, it's a gray area in the law. The software is release under the MIT OSS license and the Read Me contains the disclaimer "This is alpha-quality software for research purposes only... You are responsible for complying with local laws and regulatons." The U.S. Supreme Court, in a series of court cases in the 1990s, ruled open source code as free speech protected under the First Amendment of the U.S. Constitution.
The question is does that release the author(s) from liability. The EU has no EU wide rules on liability in such cases. One open question is even if the person who used the software could not sue, a third party injured by it might be able to since they are not a party to the license agreement.
An EFF attorney told HotHardware "Prosecutors and plaintiffs often urge courts to disregard traditional First Amendment protections in the case of software." But not everyone agrees. "Most legal experts that spoke with IEEE Spectrum -- and Hotz himself -- believe that if you use the company's code and something goes wrong, then it isn't liable for damages. You are."
The question is does that release the author(s) from liability. The EU has no EU wide rules on liability in such cases. One open question is even if the person who used the software could not sue, a third party injured by it might be able to since they are not a party to the license agreement.
An EFF attorney told HotHardware "Prosecutors and plaintiffs often urge courts to disregard traditional First Amendment protections in the case of software." But not everyone agrees. "Most legal experts that spoke with IEEE Spectrum -- and Hotz himself -- believe that if you use the company's code and something goes wrong, then it isn't liable for damages. You are."
Re: (Score:2)
It's coming anyway - just like bridge builders and other REAL engineers.
No, software engineering is not just like bridge building. Only simple software can be proven correct - bridge builders have equations they can use to verify their work.
Re: (Score:2)
Neither bridge building nor software engineering have mathematically proven models. You can approximate and calculate a lot of it based on years of construction but you still have to have common sense and experience.
The reason we have shitty software is because nobody typically dies when something goes wrong (and in the cases where it does, the instances are so rare the companies behind them often don't care enough). Bridges on the other hand are not nearly as lean and cost effective as they theoretically c
Re: (Score:3)
> The reason we have shitty software is because nobody typically dies when something goes wrong
The reason we have shitty software is because nobody wants to pay for having well tested software. The costs would be factors higher.
Re: (Score:2)
Re: (Score:2)
I'd advocate for liability for the companies that employ the engineers. In construction, the engineers are often not directly responsible unless you can prove intent, although they may lose a license in cases of severe incompetence. The construction company may be on the line for some of the construction costs, but often the governments involved ends up paying for those sorts of losses. In most cases however, it's not necessarily the engineers but the construction company and/or their subcontractors trying
Re: (Score:2)
I'd advocate for liability for the companies that employ the engineers. In construction, the engineers are often not directly responsible unless you can prove intent, although they may lose a license in cases of severe incompetence. The construction company may be on the line for some of the construction costs, but often the governments involved ends up paying for those sorts of losses. In most cases however, it's not necessarily the engineers but the construction company and/or their subcontractors trying to squeeze out a few dollars in the process that are the problem.
AFAIK, it doesn't take *severe* incompetence for structural/mechanical/civil/electrical Engineers to be sanctioned or potentially lose their license. Simply not following standard engineering practice or signing off something which is outside your area of practice (which often implies culpable negligence) is enough.
Companies can't simply hire "engineers" to perform engineering in most fields, they often have to hire Engineering companies whose principals are actually licensed Engineers, or if they employ c
Re: (Score:2)
It's coming anyway - just like bridge builders and other REAL engineers.
So what's that got to do with Open Source? They will either continue releasing with licenses that explicitly avoid any claims for warranty or liability just like in the case we are talking about right here [github.com] or it will just kill Open Source because nobody wants to open themselves up to that sort of legal backlash just to release OSS.
Re: (Score:2)
Jesus I don't even know where to get started with this.
Docker, Python, or non-RTOS.
I clicked on the link hoping to see maybe some Simulink model that compiled to a XPC or other embedded hardware. How is linux supposed to know to ignore the ethernet driver and concentrate on running the steering wheel?
Re: (Score:3)
It's easy with systemd. You just have to write it in the form of a Sanskrit poem.
Re: (Score:2)
Python is single-threaded and thus it's speed and results should be predictable. Not sure whether Python would still run on a decent real-time processor though (x86 has not been in a long time) because it's overhead would be enormous.
Re: Python? (Score:2)
Wow.. I'm impressed. Almost everything you said it's false!
Python explicitly runs as a single thread (thanks to the GIL partially) and uses time slicing to simulate multi threading.
The is also no such thing as a real time processor. I guess you think you mean a real time os? But I'm guessing the.
And no. Python performance is far far from predictable on any os it commonly runs on.
Nice language for UI development and glue code though.. use it a lot. But most certainly not hard real time.
Re: (Score:2)
I don't know much about Python since I don't use it. I know it's single threaded because I tried multithreading some existing code once and gave up.
A well-designed single threaded program should be predictable after compilation, what the hell is it doing that makes it non-deterministic, doing a "sleep rand(1...10)" after every line?
I meant a processor that you can predict it's state after every line, for most small microcontrollers you can easily (even manually) trace every code interaction and predict how
Re: (Score:2)
One example is garbage collection.
Re: (Score:2)
... which Python only uses for auxiliary purposes. Just call gc.disable(). Poof, no garbage collection. No memory leaks as long as you do not use data structures with reference cycles, either.
Re: (Score:2)
I know it's single threaded because I tried multithreading some existing code once and gave up.
Oh well if you tried and gave up then it must be single-threaded then right? In fact you can do multi-threading in Python - regardless of whether you gave up trying or not - and generally time-slicing is used to run the different threads "concurrently" on implementations that rely on the global interpreter lock.
Re: (Score:2)
Python IS single-threaded, multi-threading is a big hack on Python and usually (at least a few years ago) came with big warning letters not to use unless you absolutely knew what you were doing.
Time slicing is not multi-threading, can it allocate a process to a particular CPU or guarantee it's still running on the same CPU when I call back? I think that's the same issue Rust has.
Re: (Score:2)
Python IS single-threaded, multi-threading is a big hack on Python and usually (at least a few years ago) came with big warning letters not to use unless you absolutely knew what you were doing.
No, the Python interpreter is single-threaded, that doesn't meant that things like I/O operations have to block everything else, in fact anything that isn't being interpreted can be done outside of the interpreter lock.
Time slicing is not multi-threading
I didn't say it was, I said that is how (on some Python implementations) concurrency is implemented, much like the way you can multi-task and run multi-threaded processes on a single core CPU.
Re: (Score:2)
Re: (Score:2)
Let's assume you're talking about CPython, because Python is a language, not an implementation.
No it doesn't. CPython supports threading.
No it doesn't. CPython uses OS threads, it does not do its own time slicing.
What you're thinking about is the GIL, which ensures that only one (real) thread is running *inside the interpreter* at any one point. You can spawn multiple CPython threads and they will be *real* threads sc
Re: (Score:2)
Where is the 'Cowboy Neal' option?
Cowboy Neal is your co-pilot?
The User is responsible for open sourse software (Score:5, Interesting)
If you download, compile (or trust some 3rd party build), and use you are responsible.
If your self driving car, using open source software, is in an accident, you are responsible. There is no responsible 3rd party.
This is old law.
This is the meaning of use at your own risk.
Accept it, or don't use it.
Re:The User is responsible for open sourse softwar (Score:5, Insightful)
I have to agree here.
If I buy a part for my car and the part's manufacturer claims that it complies with some ASE or similar standard, then if the part fails I might have a legal case (e.g., if I can prove negligence in the design or manufacture, or something that the established case law will respect). However, if I buy a part off a guy who makes them in his tool shed and hey tells me "hey, I'm not sure that this thing won't explode when apply the brake," then I am pretty sure I have no recourse whatsoever.
How is software different? If the manufacturer warrants it, then it should work as it is represented and if it fails then there is a discussion to be had. If the manufacturer disclaims warranty and it breaks (and the applicable laws don't override that; you know that in some jurisdictions that there are laws that still hold the maker or seller responsible to a degree for things they make or sell?) then I don't have a legal case.
Of course, even professionally produced commercial software normally has a EULA with a clause that reads something like "the manufacturer provides no warranty of merchantability or fitness for a particular purpose and shall not be liable for losses arising from blah, blah, blah..." If you are NASA and paying your contractors (an enormous amount of money) to mathematically prove their software correct then you might get a "yup, we certify that this software will work as designed," otherwise you have no such assurance.
Re: (Score:2)
Re: (Score:2)
Just last year an 18 wheeler made a careless lane change and scraped the bumper of my wife's car. (It was a peak hour traffic at 4 mph or so). It was the trucking company's insurance that paid for the 500$ damage.
Limits (Score:2)
if I buy a part off a guy who makes them in his tool shed and hey tells me "hey, I'm not sure that this thing won't explode when apply the brake," then I am pretty sure I have no recourse whatsoever.
There are limits to this: if the seller in question made the brake pads out of plastic explosive I'm pretty sure the police will soon be knocking on his door. Putting a disclaimer on things does not magically allow you to get away with anything and some countries like the EU have mandatory minimum guarantees.
However in this case the software is not sold but given away so there is no sale which probably keeps the author protected unless they put something deliberately bad in the code.
Re: (Score:1)
Parent is 100% wrong. The reason you are typically responsible for open source software is because the license comes with a disclaimer of warranty and of liability. For example, the GPL [gnu.org] license says:
15. Disclaimer of Warranty.
THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM “AS IS” WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION.
16. Limitation of Liability.
IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MODIFIES AND/OR CONVEYS THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.
This does not prevent a company from selling you a warranty for a fee. e.g. Enterprise software customers get a warranty, because they pay for it. If you're buying a self driving car, insist on a warranty.
tl;dr: Parent is wrong; do not buy a self-driving car that doesn't come with a warranty.
Re: (Score:2)
Re: (Score:2)
So what you are saying is, al queda and ISIS have to simply open source their bomb making recipes and release it under GPL, then they will be free of liability. Right?
Nah, they don't even need to do that. Publishing is not a problem in most of the world. Even building and posessing them isn't a problem if you follow the law and basic safety rules. People and companies use explosives that could be used as bombs all the time.
If you're talking about publication only, The Anarchist Cookbook, first published back in 1971, describes how to make all kinds of bombs, explosives, and poisons, as well as assorted drugs like LSD. It is still in print, and pirated versions are av
Re: (Score:2)
The letter of the law is sort of unimportant. If you thought Ford was going to be responsible, then it is not your understanding of law I would call into question first, but your basic common sense. A car would practically have to cost twice as much, if the manufacturer was responsible for all accidents and deaths it caused.
Re: (Score:2)
Much like pretty much every open source license the one here is no different. It's open source, you have the freedom to do whatever you like with it, but you are the one who assumes responsibility for it too.
Copyright (c) 2016, Comma.ai, Inc.
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publi
Open source (Score:2)
Re: (Score:2)
Whomever installs the software is going to be on the hook.
Re: (Score:2)
The installer/maintainer (whoever you paid to 'make it work'), the source code license doesn't matter really in these kinds of things. Just because I make the code to a pacemaker open source and it fails, doesn't mean that I'm not liable. That would be a very easy way to avoid liability though.
Re: Open source (Score:2)
You do realise in this context installer and maintainer are two very very very different things . Right?
This whole article is retarded.
Of course it is the person who decided to let it control the car. Black and white. No gray area at all.
In exactly the same way they would be responsible if they decided to fit cheese for brake pads.. Or tint their front window with house paint.
The disclaimer didn't matter. The fact is there is no CLAIM that it is suitable for normal road use.
Re: (Score:2)
Not necessarily the person that wanted it, the person that ended up installing it in contravention of rules and safety. I meant the installer that mechanically connects the drive shaft and the brakes to this machine. The maintainer of the software has nothing to do with it in this instance he's just putting his science project on the net, unless he sells it (and/or gives it away) with some sort of guarantee that it will work.
Once you have a company and you sell it to do some sort of function and it ends up
Re: (Score:2)
Re: (Score:2)
With the license attached to it, the software maker has given no guarantee that this will work for any purpose. Most open source licenses don't guarantee the product to work. This is where proper engineering comes into play (including mechanical etc), if there is someone (a company perhaps) that wants to commercialize it, then they'll be on the hook for any faults, problems, testing, licensing.
Obviously at some point, failures will happen just like current cars fail, who is on the hook depends on whose faul
Re: (Score:2)
Could be like airplanes (Score:2, Insightful)
Sure, the store may go after Joe the installer, but that is their problem, not yours and the re
As it should Be (Score:2)
If you drive a car over a carload of nuns, you're liable. What's new here?
Re: Self driving cars are the new dogs (Score:1)
That's not true. My neighbor's dog "borrowed" his car one Sunday last year and ran it into the front of the local pet store. Made off with two bags of premium dog food, three cats and $340 from the cash register. Neither one of them got charged.
That's not hard (Score:2)
If you package something for sale, you are responsible. Here's a box filled with stuff for 49.99, oops the box exploded one or 10 times. The boxing vendor is responsible, regardless of the interaction that caused it. There is no specific law, but tort law generally plays out that way. A separate boxer? Not that guy's problem but he always has to go to court to be found, not at fault. Nothing here about software liability applies excepting between the packaging vendor and the development team. Will we see th
Whoever (Score:2)
Whoever the courts/governments decide is responsible on a case by case basis, or course!
What about open source bomb making software? (Score:2)
Begging the question (Score:2)
This begs the question (assumes it and ignores it): When a NON-open-source software program is involved in an accident, the responsibility is that of the manufacturer.
That is not true according to current cases dealing specifically with Tesla.
A better question isn't "Hey if an open-source independent vehicle software program causes a crash, who's responsible" but rather: 'Who is responsible when software causes a crash" or better yet "How can people be responsible for their own behavior even if relying on
Re: (Score:2)
That is not true according to current cases dealing specifically with Tesla.
Considering it triggered a federal investigation that cleared Tesla I would say the exact opposite, the current case established that the non-open-source vendor was potentially liable depending on the error.
In the Tesla case the only reason they were let-off was because Tesla's legal team required numerous liability warnings that are enforced by the software and the situation was sufficiently challenging that it would be unreasonable to expect the software to handle the edge case.
All of this though is poin
Re: (Score:2)
Answer: God. (Score:1)
God is responsible, die suckers.
Who? No one. (Score:3)
Oh, and by the way? The way I think this should be handled legally-speaking, since it's obvious we'll be subjected to these gods-be-damned things regardless, is the same way aircraft mechanics are treated, legally: If a plane they worked on crashes and people die, and the cause of the crash is proven to be a mechanical failure that's the responsibility of the mechanic that worked on it, he is arrested and tried for murder. A so-called 'self-driving car' kills someone? The programmer(s) responsible for not doing their job correctly get thrown in jail charged with murder. Oh and before any of you give me shit for this? You all make a big point about how 'self driving cars will save lives'; well if they TAKE a life then some HUMAN has to be held criminally responsible for it, plain and simple. Otherwise you're just hypocrites.
Re: (Score:3)
Re: (Score:2)
Looking at all the comments following this, you are apparently wrong. If fewer people are killed, then, it must be a loss for you to be wrong. Clearly, then, acording to the reasoning of Slashdot, we should strive for larger numbers of fatal traffic accidents.
ASDA (Score:2)
Seems simple enough (Score:3)
The owner of the vehicle is responsible for ensuring his vehicle is safe to use. If he modifies it, by installing some untested software, he is most certainly responsible for the consequences if he then injures someone. And if the disclaimers are clear enough, his chances of successfully suing the software developer are slim.
The sad thing is that, as with the Tesla, you know some idiots will install this and go on to kill people - hopefully just themselves. After that I wouldn't be surprised to see the software developer sued claiming the warnings on the software were just not clear enough. Or even the car manufacturer, for allowing the vehicle to be so modified in the first place...
that clause.... (Score:1)
"The software is provided "As is", without warranty of any kind, express or implied, including but not limited to the warranties of merchantability, fitness for a particular purpose and noninfringement. "
This is from the MIT license, I'm sure other open-source licenses have a similar clause. It basically tells you that using this software, is at your own risk.
State is responsible for OpenSource Software in ve (Score:1)
The principle is that the state are responsible for ensuring the safety of the vehicle So if a vehicle have been certified safe by the authorities they take responsibility for the safety. On the other hand if there is no regulations regarding software used in cars or that software is proprietary then the manufaturer needs to take responsibility.
In the case of OpenSource there is definitely a possibility that the owner might be responsible. But American law is funny so it could go both ways.
It's not about open source or not (Score:2)
It is about who declares to take responsiblity.
Someone (i.e. a auto company) can create an open software, release a signed version and take the resposiblity (to a defined degree). If they don't, nobody actually needs to software, as you cannot use it without driving yourself, or your car is driving illegally.
If you now use this software, the liability is clear. When you start modifying it, it becomes interesting. Because either you have the full liability for the selfdriving part yourself (which probably me
Insert Open Source licensing FUD (Score:2)
Nobody is liable going by the Microsoft Windows 10 EULA [microsoft.com]. Indeed the license specifically bars you from sueing them in a court of law and even then you can only get back what you paid for the software or $50.
"One open question is even if the person who used the software could not sue, a third party injured by it might be able to since they are not a party to the license agreement."
The third party can't sue the first party precicely because the first
Python (Score:2)
The AI left plenty of white space. But other drivers kept pulling into it.
the renter of the car who you don't own is! (Score:2)
the renter of the car who you don't own is! Even when it's just an auto taxi I hope you don't drop the soap after the auto taxi BSOD and kills some one.
Re: THE DRIVER (Score:1)
How much for death how much for loss of limb
Re: (Score:2)
Any insurance company will take a long and hard look at the risk involved. What are the odds of an accident happening with any particular flavour of
Re: (Score:2)
That's the problem, the "driver is responsible" model is useless. Babysitting a self driving vehicle is no easier than just driving it. The untapped resource with self driving vehicles are the state and federal DOT's. With fairly cheap roadside infrastructure (including cameras, centimeter accurate local positioning systems, vehicle location recording and broadcast) and a certification process for what software must do to drive a car, responsibility for the whole system becomes distributed over companies AN
Re: (Score:3)
Actually, It's me. Sorry in advance for all the dead people, all of that is my fault. Even the ones ain't died quite yet. Although I do feel badly, I do. Again, sorry.
Re: (Score:2)
So the driver is responsible. But since the human isn't driving the self-driving car, the car itself is responsible?
Its your fault for getting in the auto pilot death machine and running over those babies.