GitHub accused of varying Copilot output to avoid copyright allegations

GitHub is alleged to have tuned its Copilot programming assistant to generate slight variations of ingested training code to prevent output from being flagged as a direct copy of licensed software. This assertion appeared on Thursday in the amended complaint [PDF] against Microsoft, GitHub, and OpenAI over Copilot's documented …

[that one in the corner]

Credible threats of violence directed at their attorney

Presumably from the script kiddies and wannabes who realise they'll never make it as programmers without every single ounce of spoon-feeding they can get from Copilot.

Whatever happened to the shy nerds who used code as their means of expression?

[b0llchit]

Re: Credible threats of violence directed at their attorney

Whatever happened to the shy nerds who used code as their means of expression?

They programmed the copilot and are now sitting in their basements awaiting their code to finish writing itself, as their new means of expression.

[Anonymous Coward]

Re: Credible threats of violence directed at their attorney

Did they also program OpenAI to file complaints on their behalf?

"“go f*****g cry about github you f*****g piece of s**t n****r, I hope your throat gets cut open and every single family member of you is burnt to death.”

Sure sounds like they hate + racist personality AI has.

[Anonymous Coward]

Programming is dead, long live development

The real talent all got displaced somewhere around the time computer programming morphed into what we call software development today. The opportunists (playground bullies of yore) moved in the moment they realised they could take, take, take from the original works of truly passionate nerds (to line their own pockets) without having to learn anything about how computers work. The most disturbing thing about our current trend is that society is now heaping praise upon those who bruteforce solutions to problems, instead of championing people synthesising knowledge in the name of inventing better ideas. That needs to stop.

[CapeCarl]

Alls fair in Love and Grep?

Perhaps MS should show complete CoPilot openness to ingesting any and all code protected by various licensing schemes...By feeding in the whole MS code base.

...Hmmm on second thought, maybe not.

[Anonymous Coward]

Re: Alls fair in Love and Grep?

".By feeding in the whole MS code base."

By *not* doing that they indirectly admit it's all about circumventing copyright and they don't want *theirs* to be circumvented. What a surprise, eh?

MS has *huge* code base and CoPilot definitely could use it. If nothing else than an example of 'do not do this'.

[Anonymous Coward]

Meanwhile at client...

Client: we encourage use of CoPilot and Cowhisperer! Go for it! Accelerate the codes! (and allow us to replace all senior devs with cheap junior drones out of uni)

Me: What about the copyright and licensing issues (provides links on internal forum)

Client: hurr, durr? Use Copilot! Love Copilot!

Me: Hello? <client> (with big pockets) may get sued? Meh, no skin off my back

Client: <crickets>

[Anonymous Coward]

Re: Meanwhile at client...

Amongst the clauses in one employment contract (that I didn't sign), was that I indemnify my employer for any intellectual property breaches by the company.

[Anonymous Coward]

Sleight of hand?

... in response to public criticism of Copilot, GitHub introduced a user-adjustable Copilot filter called "Suggestions matching public code" to avoid seeing software suggestions that duplicate other people's work.

How does that "help"? If somecorp doesn't check the box how does that prevent Copilot from violating licensing - Copilot is still selling code without proper licensing attribution.

Is that protected by section 230 - the same law which protects Amazon from any liability resulting from selling counterfeit, copyright violating, or dangerous products/content online?

I actually use co-pilot and didn't check the box because I was interested in seeing what kind of things popped up.

Maybe it is the kind of code of I am writing, which is not very generic, but so far (after about 2 months of usage) I have seen nothing other that what can be attributed to pattern matching auto completion.

This case being reported seems to be placing the onus the complainant to prove that code is being at least nearly copied, and I don't think that is going to be easy in either the short or long run. It would be much better - infinitely better - to place the onus the data scrapers to not access open source software (or any other material) at all unless the resulting product follows the open source licensing agreements to the letter.

That however would require the US legislature prioritizing something other than the lobbyists.

[FeepingCreature]

Re: Sleight of hand?

Looking at code to learn from it simply is not protected under copyright.

It doesn't matter if it's open-source or closed-source, except in that closed-source makes it practically difficult to look at it without illegally copying it.

Copyright only kicks in once you actually reproduce the code.

(Yes, Wine does a lot of cleanroom engineering, but I'm not convinced it's actually necessary; it's just that nobody wants to risk going to trial against MS to find out.)

[CowHorseFrog]

Instead of wasting time with co-pilot why doesnt github invest in search that actually works. For bonus points make the search context aware.

[sgp]

Because hype

[Pascal Monett]

Yeah, but GitHub is now Borkzilla'd, and Borkzilla has never known how to Search.

[ChoHag]

So their solution to having copied loads of stuff they're not allowed to is to not show it to you?

[veti]

... Yes?

No copy, no copyright infringement. There's no law against reading code on Github.

Would it be illegal if a human did it? - to my mind that's the only test that makes any kind of sense.

[JessicaRabbit]

Pretty sure it would be, yes. That's why you have to have things like clean room reverse engineering.

[Anonymous Coward]

How close does code have to be?

Two devs can produce markedly similar if not nearly identical code independently, depending on what the code is supposed to do and how small the routine is (with different labes and variable names). At what point does one have to worry about the appearance of one violating the copyright of the other?

[veti]

Re: How close does code have to be?

Good question, and one reason why copyright is not a good tool for protecting software. Even if you accept the premise that software shouldn't be copied, which personally I still have doubts about.

The thing is, software is not a creative work. It's a tool. Its value comes from what it does or what it can be used for, not from merely existing. That's a poor match for what "copyright" is meant to protect.

[Anonymous Coward]

Re: How close does code have to be?

"That's a poor match for what "copyright" is meant to protect."

It exists to protect money. No other reason. Remove copyright and you could copy any software for free, no problem. Basically same business model as open software has now: Money is in the support, not in the program itself.

The hairy part comes when copyright is streched to single lines of code or snippets of code and it does that *only* because record & TV companies wanted it to be so. No other reason. Almost all of the copyright legislation is created by Hollywood media magnates and that shows everywhere.

[that one in the corner]

Re: How close does code have to be?

> {copyright} exists to protect money. No other reason.

Stated crudely enough to be inaccurate, as I'm sure you know. Copyright exists to protect the interests of creators and allow them to make a living - purely in order that they spend time creating things - without the need for a patron. Especially as the ability to reproduce increased over time. Copyright and the controlled selling of mass reproductions allowed us to provide patronage en masse, all of (what we now refer to as mere "consumers") providing a small portion of the creators interests. And, yes, these days that means money, simply because that is a good mechanism for handling such portions: individual patronage often being in the form of goods and services provided, including lodgings.

> Almost all of the copyright legislation is created by Hollywood media magnates

The current rotten state of the copyright legislation has been caused by those magnates - more accurately, political systems that enable those magnates to affect the legislation. But that is another matter.

The concept of copyright is definitely a Good Thing. Its execution - varies.

[vekkq]

Re: How close does code have to be?

If you would sell the work, not the product, there would be no need for copyright.

[doublelayer]

Re: How close does code have to be?

Which works if you can find one person who wants the product enough to be the only customer. Otherwise, the person buying the work tends to be selling the product themselves, so just because you're not selling it directly doesn't change the situation. Basically, if you or someone else can't sell the product, you'll often be unable to sell the work that would have created the product.

[Orv]

Re: How close does code have to be?

Open source licenses like the GPL would have no protective effect without copyright; someone could take the code, use it commercially, and never publish it. This isn't just about commercial licenses.

[doublelayer]

Re: How close does code have to be?

"Remove copyright and you could copy any software for free, no problem. Basically same business model as open software has now: Money is in the support, not in the program itself."

Except for three major differences/problems:

1. That business model doesn't consistently produce the kind of profit needed to develop all the kinds of software that the copyright model does.

2. They are sometimes based on not having other companies copy their work, which works pretty well when the license is structured to require modifications to be published under the same terms and enough people are using the code to resolve forks, but not on other types of projects.

3. This would only become the same if the code was published, but "There is no more copyright protection" does not automatically lead to "All source code will be published".

[Orv]

Re: How close does code have to be?

Indeed, "no copyright protection" seems like a good way to encourage DRM systems. Just like how the alternative to patents isn't open publication, it's trade secrets.

[OhForF']

Software is not a creative work?

I'd say software is a detailed list of instructions on how to solve a specific (class of) problem(s) and imho generating that list requires creative thinking. If it doesn't require creativity it should be easy to automatize and so far attempts to automatically generate programs have not been very effective.

What is your definition of a "creative work" if it doesn't apply to software and how to you plan to fairly compensate programmers for their work if you allow free copying of everything they produce?

[veti]

Re: Software is not a creative work?

Consider the browser you're using right now. Of course it's a work of software, and the lines of code used to write it are protected by copyright. But the "lines of code" are really the least interesting thing about it. Far more important is the design - visual design, interface design, component design.

If I write a clone of your browser using all new source code, I've taken advantage of the hard work of the original designers. Yet "copyright" gets horribly bogged down at this point. Have I infringed it or not? That's a question to keep lawyers employed. (I may, of course, have infringed design rights, but that's a whole separate can o' worms.)

But it's the design that's by far the more creative aspect. Why don't we talk about some form of protection that focuses on that? - instead of obsessing about lines of code, many of which are probably indistinguishable from lines used in thousands of other applications across scores of different domains.

[doublelayer]

Re: Software is not a creative work?

"the "lines of code" are really the least interesting thing about it. Far more important is the design - visual design, interface design, component design."

I disagree. Sure, design is sometimes useful, but if Firefox moved the menus around, I could deal with that. It wouldn't be the most important aspect. Same if they changed the colors or the visuals around the controls, or changed the structure of the script parsing and execution system. What is most important is what it does and how it does it, and it is the code that defines both. Not to mention that it is also the code that implements the design. While each individual line of code is not critical to me, the thing that they allow me to do is. Whether the tracking blocking has a switch in the settings, in the menus, or its own keyboard shortcut is relatively unimportant. That there is tracking blocking and that it's been kept updated to continue to work against new exploits that advertisers try to use is very important.

The same is true of any other creative work. Take a good book. The core part of any book is its plot, setting, and characters, but this is not a good book:

Winston Smith: employee of despotic government, not having fun, doesn't like situation. Government tortures people for no good reason. Smith tries to think of a way around the absolute dictatorship, meets a fellow dissident, caught by police, tortured, ends up depressed, dictatorship as strong as it ever was.

All those connective sentences, each of which is relatively unimportant, is what turns this boring list of statements into something that's enjoyable to read. While many people wouldn't come up with a plot like this if they tried, many others could envision it but couldn't make it with the quality that Orwell could, and some people could make a book that is just as good, but takes those plot points in a different direction. Each individual sentence is relatively unimportant, but as they build up, they create something that is bigger than the original picture in the author's mind. This is why the idea of a dictatorship and a person struggling under it isn't copyrightable, but 1984 is.

[tiggity]

Re: Software is not a creative work?

@veti

Disagree.

I write a lot of APIs / microservices - providing lots of "core" functionality.

These are used by various apps - ranging from websites through to desktop applications: Calling apps can have a variety of design / UI features, but the "grunt work" is done via the same APIs

[that one in the corner]

Re: Software is not a creative work?

> instead of obsessing about lines of code, many of which are probably indistinguishable from lines used in thousands of other applications across scores of different domains.

Instead of obsessing about individual notes, many of which are probably indistinguishable from notes used in thousands of other scores across lots of different styles of music.

Instead of obsessing about groups of letters[1], many of which are probably indistinguishable from words used in thousands of other books across scores of different genres.

> visual design, interface design, component design.

> But it's the design that's by far the more creative aspect

Yikes, this is starting to sound like an argument from a "creative" who hasn't had the enjoyment of coding something new :-( But putting aside such derogatory remarks (tut tut Corner, Old boy):

Copyright only covers the *expression* of an idea/design/... and even by your own argument, that is the code (there are other ways to express the idea/... - you may have drawn your UI out on paper and written in the Pantones - and that is also copyrightable; but the code is the expression that is actually *useful* and worth disseminating).

Depending upon your location, you may be able to get a design patent on, well, the design; you may also be able to trademark some of it (e.g. a particular shade of purple used in a specific context - a list of applicable contexts is available from your local trademark office).

[1] yes, given the nlocs in something like Firefox, comparing lines of code to words in a novel or notes in a score is reasonable: certainly makes it easier to structure these comparisons!

[SloppyJesse]

Re: Software is not a creative work?

> If it doesn't require creativity it should be easy to automatize and so far attempts to automatically generate programs have not been very effective.

Of course it's easy to automatize. If you can just write down a detailed list of exactly what you want it to do...

[that one in the corner]

Re: How close does code have to be?

> software is not a creative work. It's a tool.

A *completed*[1] piece of software is a tool - we freely admit that in the documentation: compilers, linkers are part of our tool chain. Other software may be "just" a utility whilst bigger lumps are called applications.

But *making* the software and completing it or adapting it, *that* is a creative act. There are lots of stages that are trivial, repetitive and can even be totally automated[2], but the same is true for other creative pursuits: I'll sand with a dremel because it is less repetitive and tedious than using a teeny tiny sanding block[3].

> Its value comes from what it does or what it can be used for, not from merely existing

That is a trivial observation, too trite to even be a truism. Name anything that literally "has value" merely by existing! "Value" is something that we attribute on the basis of our personal preference or how we react to someone else's preference or actions with respect to it. That driftwood has no value to her because it won't burn cleanly in the hearth, but he can see how to use it in an artwork. Even money - a fiver has no innate attraction to me (the plastic ones are quite irritating sometimes) but I give it value because of how it can influence others actions (like letting walking out of the shop with a new pair of socks without angry shouts).

[1] ok, we all know that no software is ever really "complete", but you know what I mean.

[2] closing brackets, expanding a template to insert an entire loop - as time passes we attempt to increase this list, in order to spend more time on the interestingly creative bits.

[3] and my results are certainly unique! Hopefully I'll get better with practice and someone may be able it is "creative" with a straight face! It is *meant* to be a dragon, btw.

[veti]

Re: How close does code have to be?

The Eiffel Tower.

Does it have value? To many people, certainly. Given that "value" is inherently subjective, that's as much as can be said for anything.

Does its value depend on the uses to which it can be put? Well, partially perhaps. It's a tourist attraction, a metaphor, a landmark, an argument. But even if all those uses somehow dried up, the thing itself would still be dear to many people.

Nobody (for statistical values of "nobody") gets software to experience the thrill of decompiling it and enjoying the intricacies of its design. They get it in order to watch dancing pigs. Nobody gives a single flying fornication about the code used to make the pigs dance - they're too busy wondering if they can do Gangnam Style. If someone offers them a way to watch the pigs without using that piece of software at all, they'll drop it with never a backward glance.

[Michael Wojcik]

Re: How close does code have to be?

I honestly have no idea what point you're trying to make with this argument, which seems pretty naive from the perspectives of economics, psychology, and aesthetics.

Software patently has both use-value and exchange-value.

As a non-physical, non-exhaustible, non-rival good, it has no unique-object value, sure. So what? Walter Benjamin pointed out almost a century ago that such "auratic value" no applies to most creative endeavors. Welcome to modernity.

People associate the experience with the product, and the qualia engendered by the experience attach to the product. We see that with video games, for example, which are definitely software and which many people form definite attachments to that go far beyond the experience of creating the game. (I haven't looked, but I'm sure there's plenty of Aeris/Tifa slash out there.) For more mundane software, there's certainly ample evidence of people making emotional investment in their preferred packages – just as they do for other tools.

[Rol]

Re: How close does code have to be?

Coding is a highly creative process, involving a plethora of disciplines and despite two competing sets of code performing the exact same function, they can be a million miles apart in speed and efficiency.

So do we copyright the concept of sorting a list of data, or do we rightfully recognise the art of doing that as efficiently as possible and therefore the code becomes the protected art?

[Paul Hovnanian]

Re: How close does code have to be?

It would please me greatly if the rest of you would stop copying my "main(int argc, char *argv[])" opening line. Thank you.

[that one in the corner]

Re: How close does code have to be?

Please be aware that I have applied for a patent on the concept of changing your declaration of main() as follows:

main(int argc, char **argv)

Please direct all licensing enquiries to Under The Bridge Patents (Holdings) Ltd, Basingstoke.

[katrinab]

Re: How close does code have to be?

At the point where you actually copy. If two people come up with the same thing independently, it isn't copying, and therefore isn't covered by copyright.

Patents might cover it, but that is a different law. Depending on jurisdiction, the first to invent or first to file could get a patent on it, and that protects you against someone else coming up with the same idea independently.

[localzuk]

Re: How close does code have to be?

This is what I was thinking. Is one line of code copyrightable? How about 2? What amount of code, which is effectively just rewritten from the documentation of that language to achieve outcomes as documented, makes code copyrightable?

How novel does code need to be? Obviously a hello world script is not novel enough (or is it), but would a method to convert an inputted file from one format to another be novel enough?

[Michael Wojcik]

Re: How close does code have to be?

Is one line of code copyrightable? How about 2? What amount of code, which is effectively just rewritten from the documentation of that language to achieve outcomes as documented, makes code copyrightable?

Gosh, if only similar questions applied to other works which are subject to copyright, and were discussed in both statute and jurisprudence.

Oh, wait, they do, and have been. In the US, some of those cases specifically regarding software are quite famous, such as Lotus v. Borland and Oracle v. Google.

[localzuk]

Re: How close does code have to be?

Huh? First, the whole world is not the USA.

Second, Lotus v Borland does not cover this - that was about menu structure, which was determined not to be copyrightable, but did not set a national precedent...

Oracle v Google is closer to the topic but isn't what I was asking. Maybe I should have included "Should" instead of "Is"? So, should 1 line of code be copyrightable, 2 etc...

What the law says, vs what it should be are different things.

[doublelayer]

Re: How close does code have to be?

"the whole world is not the USA."

They didn't say it was. "In the US" is a dependent clause, meaning that the cases they're about to mention are from that country, not that everything is.

The precedents set so far don't cover everything, but the general concept applies to many other kinds of work and are frequently tested. Small chunks of music are probably the most direct similar cases, where a few notes in a sequence can be considered unique enough to be copyrighted, but in many cases are considered small or common enough that infringement of them is not possible. However, clips of recorded audio are also copyrightable, and those stand up to more protection. A lot of it will come down to a judge's idea of common sense, which while not being a perfect solution because it isn't deterministic, often leads to functional verdicts.

[Ashto5]

GitHub purchased by MS

Well no one saw this coming …

MS purchase the largest code repository on the planet and then scrape it to sell you code that really was not theirs to utilise.

Everyone wins IF you are a lawyer that is.

[Anonymous Coward]

".... was trained on publicly posted code in a way that violates copyright law and software licensing requirements and that it presents other people's code as its own."

It's Microsoft, that has been their mode of operation since QDos, literally *before* Day 1. Proving that in court is another matter and that's why Microsoft has a lot of very high priced lawyers.

You see, it's not theft if you can delay the case until the accuser goes bankrupt. Easy peasy. Or, in DOJ case, you get your guy as president. That's the proper MS-way to do things.

[Michael Wojcik]

It's Microsoft, that has been their mode of operation since QDos, literally *before* Day 1.

Sure, if by "literally" you mean "figuratively", and by "figuratively" you mean "actually not at all, I'm just making up some bullshit".

Microsoft (founded 1975) is handily older than QDOS (written by Paterson in 1980).

[F. Frederick Skitty]

Prior to QDOS, Gates and Allen made a living by copying code they found by dumpster diving at other software companies. Allen admitted this in his autobiography.

[bo111]

Global context of intellectual property

Would we rather allow adversarial countries to accelerate their AI > economy > military, as they do not have to fuss over copyrights? Then wonder why, or why, the western economies are slowing down. Besides every day we hear about whole code bases of large software companies stolen*. Everyone is spying on everyone. Now what?

As a believer in singularity already happening**, and the global context, copyright laws must change***. Knowledge sharing accelerates economic development. Isn't it what everyone wants in the end?

____

* Interestingly hardware tech expertise is really hard to steal (TSMC, ASML). Or Taiwanese educational system.

** Singularity in this context is a rapid acceleration of socioeconomic and scientific processes.

*** Else the West becomes the countries of lawyers, not of the economical progress.

[that one in the corner]

Re: Global context of intellectual property

Huh?

> copyright laws must change. Knowledge sharing accelerates economic development.

Are you confusing copyright and patent wrt "knowledge sharing"[1]? Although both are designed to *promote* sharing: copyright allows you to publish and set terms, including licensing, for that expression of the idea.

> Then wonder why, or why, the western economies are slowing down.

Selling access to fairy dust like the current crop of LLMs (and offshoots like Copilot) is a major enough part of western economies that restricting them is going to have a noticeable effect? Then we are truly screwed anyway. Just like the fairy tale of value in sub-prime mortgages.

> As a believer in singularity already happening**

** Singularity in this context is a the creation of at least one craft beer per annum[2]

[1] copyright only covers the specific expression of an idea/fact/knowledge not the idea/fact/knowledge itself

[2] See, we can all redefine words to suit our ourselves

[doublelayer]

Re: Global context of intellectual property

"Knowledge sharing accelerates economic development. Isn't it what everyone wants in the end?"

Knowledge sharing isn't diametrically opposed to copyright. In a few ways, copyright actually helps the availability of knowledge. The obvious example is why it exists in the first place: if there is copyright, people can sell the results of their work which means they're motivated to keep doing that work and making the results public. Sure, it's not free at the start, but you get more of it because people are attracted to the work. If you don't have copyright, then someone who cannot make their work without payment will choose to do something else instead, and you'll be denied the creative production of which they would otherwise be capable.

There's another way to look at this, though. What would happen if copyright was completely eliminated, and now you can use any code you can get your hands on without anything to hinder you? Those companies which continued to make software and want to get paid for it would decide that it is too risky to leave anything open to the customer and would resort to ever stronger software locking mechanisms to prevent people from getting access to the sensitive code. Since they would no longer be able to send scary legal letters to catch the horse when it started to bolt, they'd need some really strong locks on the barn door. When those software locks were defeated, they might focus on hardware locks, and unfortunately those have been good enough to prevent most users from getting past them; consider how many Android phones never get any alternative software installed on them because the average developer cannot bypass their locks. Those companies also now have the right to take any open source code they can find, modify it in any way they choose, and lock it away since the licenses would have lost all legal force. This would end up taking a lot of software developers and making them work on anti-copying code, burden hardware with extra protection work, and you still wouldn't get access to most of that code. Does this really sound like knowledge sharing, and do you think it would lead to meaningful economic development?

Oh, and "singularity" has a meaning and your use isn't even close to it. Look it up. We're trying to communicate here, and making up new salamanders* just causes hypothesis** and foments*** the conversation.

* definitions

** confusion

*** slows down

[martinusher]

Intellectual Property Gold Rush

Although we're fed sob stories about how Big Technology is ripping off the impecunious widow or whatever we all know that its like bank robbing -- its like robbing banks, "because that's where the money is". You're not going to get sued for copyright infringement by the music publishers until you have that hit, that vein of gold to mine, and with code its all about going after Big Tech and its inexhaustible supply of cash.

So they get thwarted. What on Earth did they expect? Anyway, all real programmers know that code's just something you put together to express an idea, to mechanize something, it really doesn't have any intrinsic value (and if we were to be brutally honest, nearly all of it is copied anyway).

[FeepingCreature]

If you offered me a filter

...to auto-mutate code that I wrote that looked too much like other code posted publically, I'd take it too. I wouldn't have to prove whether some "suspiciously" similar code was intentional or not! Avoid a lot of copyright mess to begin with.

[steelpillow]

So GitHub is come to this

How far does it have to fall before everybody walks?

[captain veg]

can we be sure...

Am I paranoid to wonder if, every time I ignore Visual Studio's code suggestion and type out something different, it doesn't then hoover up my version and post it back to the mothership as new training data?

-A.