When AI helps you code, who owns the finished product? I've been writing software for almost half a century, and my recent experiences with AI suggest that developers may soon find ourselves in a very sticky situation. I say that having started with 8085 assembly code, then moving on to C, then C++, then Java. Once the web came along I learned the three Ps: Perl, PHP and Python. …
"We’ve run into this mess at ludicrous speed, blithely unaware that using these AI-powered coding tools turns the copyright protections every software firm takes for granted into a sort of Swiss cheese of loopholes, exceptions, and issues to eventually be tested in court cases."
Who's this "we"? It's been blindingly obvious to some of us.
This post has been deleted by its author
It should have been obvious to anyone with a functioning brain cell that they don't own the code created by these tools. Furthermore, as I've often said, these "AI" tools are not intelligent in the slightest, with zero understanding of any actual meaning, so they can't possibly produce a valid solution.
The more interesting question, to my mind, is whether any of that code is already subject to copyright as a result of AI having regurgitated code it previously ingested.
Whereas you know for certain that you cannot claim copyright yourself in the code AI produced for you, you do not know for certain whether someone else can. And even if pending lawsuits determine that AI training is "fair use", I doubt that the verbatim reproduction of part of a copyright work will get a pass on the same basis.
But, whatever our problems, lawyers are going to make lots of money, so let's focus on the positives.
I think this copyright thing when it comes to AI is a bit like patent trolling.
The AI doesn't "steal" any inventions or whatever. If you for example wrote a for loop that does something to database models, you are not creating anything new nor ground breaking. There is probably thousands of identical functions independently created by others.
If you browse open source code, you can find snippets of functionally very much the same code, just variables are called slightly different, formatting style is different etc. but it is the same.
If I create a hello world app, push it to Github and then ask LLM to create hello world app for me and I see it is the same as the one I created, do I have a case?
I think the problem is that you can't be sure.
Whereas boilerplate code may not be an issue in reality, the more specific the problem to which you're seeking a solution the more likely it is the AI will hone in on a specific result rather than a synthesis of results. It may well be that specific result has a permissive licence in any case - but AI won't tell you that so the mechanism that exists to resolve this uncertainty is unavailable.
A significant chunk of the economy is based around notions of intellectual property and its value. As the courts have already decided that artificial intellectual property has zero value, there will inevitably be considerable interest in preventing the former being transformed into the latter.
PS: Not my downvote.
Every code is very much boilerplate. Everything can be done in only so many ways.
The problem with copyright here, is that everyone wants to claim rights to common sentences, if I could use language analogy.
I could agree if what LLM is implementing is a process with an active patent (that is not lodged in a jurisdiction that waves everything through). Perhaps that may be a gray area, but anything else I don't see a case.
"Every code is very much boilerplate. Everything can be done in only so many ways."
I suspect the inventors and owners of compression or encryption algorithms would disagree. Fortunately for us they are mostly academics who allow free use, mostly. Some still remember GIF compression licensing. The evil geniuses behind the TikTok recommendation algorithm might be peeved if a LLM spit out an exact copy. As a developer myself I am aware that I am usually standing on the shoulders of giants, who had an amazing idea one day that turned a chore into a few lines of code. Now that genius idea is just a framework function call away, so it's easy to forget the debt that's owed. My boilerplate code relies on hundreds of previous genius ideas.
I suspect the inventors and owners of compression or encryption algorithms would disagree
The issue here is you cannot copyright or patent an algorithm. You can copyright your implementation code that you created for the algorithm. However if some one else cleanroom's a new implementation of the algorithm then that is that.
Software cannot be patented as it is just considered as the textual implementation of an algorithm. It is like a book, it can only be copyrighted.
A lot of code uses the same structures, but that is not the same as boiler plate. My for loop may look a lot like your for loop, but that does not make them the same. The loop condition line is not distinctive. The loop body easily could be. Even if it isn't, the function that contains the loop may be distinctive. Even if that isn't, the class the function in may be. Even if it isn't, [loop continues ten more times].
This is why companies have to hire programmers to build things for them. If it was all boiler plate, then they could use one of the low-code tools where you drag some blocks together and out pops an application. Many people do successfully build systems around that kind of thing. Many others try to and find that there is a situation where they need original code written. Just as each sentence in a book is not particularly special but the book ends up containing things that were not previously available, programs are made up of a small set of familiar functions and are still quite original.
"If it was all boiler plate, then they could use one of the low-code tools where you drag some blocks together and out pops an application."
This sounds like why one would use AI to create an application. The downside is that if you aren't doing most of the coding, you don't have the insight into how the program gets from input to output. For most values of input you could wind up with an accurate output, but for the edge cases you could wind up with very inaccurate output which might be dangerous depending on what is relying on those answers. A current analogy is autonomous driving. If the system is works well in many cases, the user might be lulled into a false sense of trust up to the point where the car finds it's ideal mate, something with bright red flashing lights, and tries to mate with it at high speed. Do the programmers understand why that off-nominal case happens or was a bunch of the programming washed through some automated black box?
And any of us who've been doing this for more than a few years have seen the results of an overconfident amateur (and BEEN that overconfident amateur) mashing together boilderplate and copy-pastes from various blogs and forums until something does what it's supposed to. Mostly.
Then the business case changes and they modify the code with thoughtless and expeident copy-pasting.
And seven years later management is asking why the system is crashing constantly, consuming three times more user licenses than the number of people using the system, and the consultant brought in to fix the mess is bleeding from their eyes and screaming.
> just variables are called slightly different, formatting style is different
I did some binary reverse engineering on a 64KiB bootrom in a RISC-V SoC (using Ghidra). And was able to traceback at least one tiny fragment of the binary to source code that is GPL licensed (~10 lines of source code). If they had done as you suggested the binary file would still produce roughly the same binary code.
Not sure what you mean. Compilers may optimise functionally same but visually different code to the same binary. The fact that something matches GPL code is meaningless, because that same GPL code may match differently licensed code somewhere else.
It's like saying that in your comment "If they had done as you suggested" sentence, you basically lifted it from Call The Midwife Season 12 Episode 4. You have no respect for intellectual property!
"It's like saying that in your comment "If they had done as you suggested" sentence, you basically lifted it from Call The Midwife Season 12 Episode 4. You have no respect for intellectual property!"
That's why, in cases like this, there is a de minimis test. How many books start with "It was a dark and stormy night" and go on to tell a very different tale? Some books I've read are broadly the same as others with the character and place names changes along with an altered plot twist or two. Copyright doesn't extend to ideas which is good or my library of Space Operas would take up far less shelf space and I'd have to keep buying genres I care for as much to satisfy the elephant's child.
"If I create a hello world app, push it to Github and then ask LLM to create hello world app for me and I see it is the same as the one I created, do I have a case?"
Something as simple as a "hello world" bit of coding likely isn't complex enough for a copyright claim. I'm not saying you have to have an application that competes with Photoshop before you might have a claim, but for something very simple the odds a judge/jury would not uphold your assertion is too high to spend money on blood sucking lawyers. You'd want 90%+ odds that you'd win your case so you minimize the chance you'd lose your house/car/retirement funds. Do you feel lucky?
"whether any of that code is already subject to copyright as a result of AI having regurgitated code it previously ingested"
Yes, absolutely. BUT
- Is it at all possible, given some source code written by AI, to definitively say "this code is coming from such-and-such copyrighted source"? (I would say, probably not, at least not to a legal standard of certainty)
- given that uncertainty, what is the default assumption? ie. is the onus on the original copyright holder to prove the code is theirs?
- Is it even possible, given some source code, to definitively say "this part is human written" or "this part is AI-written"? (I would say, probably not, at least not to a legal standard of certainty). It's probably highly likely that a skilled coder can reformat / refactor AI-generated code to keep the same functionality while looking like their style.
- given that uncertainty, what is the default assumption? Does a coder somehow have to prove all of their code is personally written? Or can anyone who copied their code do so with impunity and use "it's AI-written" as a blanket defence?
"lawyers are going to make lots of money"
seems like a certainty!!
Probably neither as it is too short and required little creativity.
That said, copyright is not patent. Copyright covers concrete expressions of ideas, copying and derived works. Provenance matters much more here. Two programmers can independently(!) write very similar code and both have copyright to their respective works. That is just how it is.
"The more interesting question, to my mind, is whether any of that code is already subject to copyright as a result of AI having regurgitated code it previously ingested."
There's a "compilation" copyright so if your application uses libraries and code bases that you purchased/licensed from somebody else, you can still claim a copyright on the entirety of the code for your application. It's like a book where the photos are licensed to the author/publisher for use in the book. The publisher of the book can't pursue a copyright claim against somebody that infringes on the photos but can if somebody copies the book or pages from the book and uses them in a way that isn't fair use (eg: criticism/commentary).
This post has been deleted by its author
If you write something in a high level language you would expect to own the code. However to use it it needs compiling and with the complexity and optimization heurustics of modern compilers you may not really understand the final code ... and it's quite possible that some of those ideas that gives the compilation came from compiler writers analyzing the results on someone else's code.
So maybe the AI generated python is just an intermediate stage in the compilation of the very high level language "code" you give to the AI tool?
Some of the code that comes out of a C compiler came straight from the C compiler. This includes the start up code and the intro to non-trivial functions. GCC comes with explicit licenses for that code so anyone can use GCC to create non-GPL binaries from their own source code. If a compiler does not come with such a license then its only value is to trap people who do not check what the license permits before selling the resulting binaries.
But you do know what rights you have to it before you use it. If I go online and find that there is a library, but it is commercial and I don't want to buy it, I can write my own or keep searching. I can do that because I know the terms. If I do not know the terms before using something, then it becomes harder to decide whether to use something or not.
"Now we have gen-AI as another tool for shit processes. Woohoo."
It's a tool just like a hammer. You can build shit houses or very nice houses with that hammer. If you use AI tools to help you get to a work that is good, it's a valid tool to use. Just because so much programming is a complete train wreck doesn't enter into it. Human coders just create that trash more slowly and at a higher cost.
Does the AI understand and respect the licenses of the code that it bases its suggestions on?
I can’t use third party code in my commercial environment unless the license allows me to.
Even then my organisation has strict rules about importing third party code modules. So if the AI has lifted something almost verbatim from a public repository and I lightly preen it before committing it internally then am I compliant? I don't even have the information to answer that question.
And of course there’s the moral argument of helping yourself to someone else’s intellectual output. Where physical goods are concerned there’s a word for that. Theft.
"Does the AI understand and respect the licenses of the code that it bases its suggestions on?"
No, for two reasons. A) the software is not intelligent and does not automatically discover and enforce licenses, and B) the companies that build it want to have as much training data as they can, so even if it could, they would prevent it from doing so for the same reason and in the same way that they knowingly use lots of data they don't have the rights to use.
"Did I understand the problem I wanted to solve? Could I express it clearly? Could I communicate my understanding to the AI in a sufficiently direct and unambiguous prompt that it would generate the response I sought?"
I would jolly well hope so, whether the third point referred to prompting AI or merely to documenting the code. One of the huge problems with current software development is that these three absolute prerequisites for adequate engineering are no longer considered important. Instead, we 'design' as we code, which leads to the incoherent messes that dominate the product space. Nobody would build even an office block like this, so why do we tolerate it when building software?
Really. Most of this article is just a long-winded way of saying "I have normally used terrible practices in creating software, and this new terrible tool requires different practices". And the value of those different practices will neatly be negated by using them in conjunction with that terrible tool.
Show me an AI-written story that people would be willing to pay for. Not counting flooding Amazon with tens of thousands of worthless books knowing some might get sold when someone makes a random choice. Doesn't matter that OpenAI "used" all of John Grisham's books - it still cannot create a best seller like John Grisham. It can only try to mimic his prose style.
I've read though some AI generated articles - only because they are so dumb it was kind of laughable.
The author makes clear that the author provided the essential design of the program - the assistant didn't try to correct his initial design miss.
I have seen a comment that about 90% of all code written is never used outside of the organisation that wrote it. Ie it is not sold (commercial closed source license) or open sourced or something.
If it is not distributed outside the organisation then does it matter who owns it ? You might be worried that the owner of code ingested by the AI will come after you -- but how will they know that you have used the code?
If you want to sell (a license to it) it: you cannot claim ownership. However since no one sees such commercial (closed source) code, no one will know - just as bits of open source within commercial code are only rarely discovered - vendors find it profitable to just ignore the problem.
If you want to open source what you are writing then you have the biggest problem: others can see the code and may be able to identify what the AI ingested. So if your intent is to open source then maybe what you should do is to use the AI to produce some code, inspect that code (ie learn how it works) and then write code based on your new understanding. This is much how those who write closed source code should use open source code -- except that often they just copy it.
So: those who have the most to worry about AI generated code are those who write open source code.
"Should your current employer, who sends you on various training courses, get a slice of your future income after you have left for a new job ?"
This is why some of them will, if they pay for training, require you to either stay for a certain period or reimburse them for the training. There are some situations where this makes sense and many others where it doesn't, but just like the teachers, this is already handled and does not make a good parallel to use of LLMs.
If I use "6502 Assembly Language Subroutines, by Leventhal" (to take the first book on looking at the bookshelf) to help me code, do I own the finished product?
I've been using Cody & Waite recently to code some low-level trig/log code. Do I own the resultant code?
My God! I used a Remmington typewriter to write my first up my school Technologies project. Do I own the text I wrote?????
Why do people insist on coming up with these irrelevant questions they know the answer to. You clearly know the ways to answer the questions in all your examples, and you should also know by now that this does not mean that you own the output of an LLM you use. You should know that using someone else's code from a book does not mean you own their code but that the book likely gives you permission to use it, and you do own the code you wrote around them. Using a tool to get things from your brain onto the computer or paper in this case are completely different.
According to the copyright notice in the 1982 edition of Leventhal & Saville's '6502 Assembly Language Subroutines', the program listings may be "entered, stored, and executed in a computer system, but they may not be reproduced for publication".
The licence for "Numerical recipes in C" (1988-95) is much more restrictive. It allows only transfer of sample code to a single computer for each diskette purchased, which implies that code examples can not be incorporated into wider applications without express permission (although code transcribed manually from the text is not specifically mentioned).
soon learned that I'd have to adapt my playful coding approach to something more rigorous. Did I understand the problem I wanted to solve? Could I express it clearly? Could I communicate my understanding
Your (athough probably not now extant) lecturer in Systems Analysis is probably thinking FINALLY!
Once you understand a problem well enough to effectively and clearly communicate its requirements, AI is largely irrelevant. Invariably the structure of the problem defines the solution or at least constrains the feasible solutions. These solution(s) pretty much imply the classes at algorithms that could be used to realize them.
There are a surprisingly small number of basic or fundamental algorithms applicable to conventional computing. The explosion of implementions of these basic algorithms generally reflects the specialization of data structures and methods to efficiently implement the algorithm for a particular problem-solution case.
Modern coding environments like Python have a rich set of data structures and operations which facilitate the implementation of quite high level algorithms. Most also have large libraries of algorithms and further data structures.
As for the intellectual property question I think the author's employer's opinion that you own the copyright over the compilation is pretty much on the money.
This would bring the question more in line with literary works (fair dealing and all. :) Two murder mysteries novels by separate authors (say Margery Allingham and Dorothy L.Sayers) might use exactly the same plot devices (they don't) but would still be distinct non infringing works.
Personally I have always thought a lot less IP: copyright and patents in the software arena would have hastened development in all areas of IT.
It's not coincidental that the internet explosion occurred because of a critical mass of open source, largely unencumbered software (and to a much lesser extent hardware) and not in spite of it.
Any non trivial application would have a reasonably large number of possible implemention choices which should dearly bring it under the compilation copyright head without arguing about the implementation of a particular algorithm in one small part of the application.
Otherwise trade secret protection might be more appropriate.
The reported concerns are what arise when people who insist upon the anachronistic notion of 'intellectual property' (IP) being analogous to physical property are confronted by the digital paradigm and its consequences.
Law pertaining to IP is a tangled web of inconsistencies, special provisions to suit major players in IP (e.g. Disney), and a cash cow for lawyers supporting people who trade in so-called 'rights'. The notion of IP was always specious: the inception of the digital era is bringing that truth home to people in droves.
The body of IP law is bizarrely complicated compared to that concerning 'rights' over individual instantiations of physical property. Ordinary citizens have a grasp on where they stand regarding their physical property and that of other people. In general, that perception is absent for IP. Even people active in IP related industry and commerce rely on the not always unambiguous 'say so' of highly specialised lawyers.
Moreover, market-economics has been distorted to accommodate IP. A 'product' lacking tangible existences in time and space, one devoid of natural scarcity, one for which 'price discovery' cannot hold, has been shoehorned into markets through the ploy of artificially constructed monopoly.; oddly, monopoly in other circumstances is frowned upon.
Global society and its economic underpinning is currently in a state of flux. This in large part a result of ease of communication (personal, financial, diplomatic, etc) brought on by the Internet. The so-called 'Global South' is developing muscle. It should not be taken that assumptions enshrined in international law whilst most of the Global South had colonial status shall carry forth.
IP cannot fit into market economics. What does exist is a market for people, and groups thereof, to offer their creative skills, but with no pretence of ownership thereafter of what they produce. 'Rights' cease to be an ersatz commodity. They are replaced by a general entitlement to attribution, coupled with protection against usurpation of reputation by other people fraudulently angling to gain commissions.
I share your experience and relate to it. Almost three decades ago I learned from GW-Basic, Assembly, C and now mostly Python. Over the years, from an engineer, I have become an entrepreneur and now a consultant, hardly writing any code unless I have to. But nowadays, with LLMs, I am writing again more code and yes it is Python. Instead of turning to search for an utility, I just get ChatGPT to generate a Python script to do whatever I am trying to automate during my day and life is good. Productivity has certainly improved. It is all in the prompt and my experience, you need to do about half-a-dozen interactions with ChatGPT before you can coherce it into generating something that you can use and solve your problem.
Yes, the legal question is very valid. I suppose there is no point in copyright for software anymore, other than for proprietory techniques and patents scenarios. Thanks.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
