Mise à jour
Mise à jour de la base de données, veuillez patienter...
Legal Corner: The threshold of originality for copyrightable source code
As a general rule, software source code is protected by copyright by default. Copyright holders create Free Software by applying a Free Software license to their code. Creative works, such as source code, nevertheless must be sufficiently original in nature before they can be protected by copyright. Exactly what is this “threshold of originality”, and are there types of code that are not automatically protected by copyright?
A Brief Introduction to Copyright in the European Union
Copyright is a legal construct that grants a person exclusive rights over a creative work. This can refer to almost any output/work produced by a creator: written text, a piece of music, a film, etc. Source code is included in this bundle of creative works that are protected by copyright. In the European Union, this was confirmed by the European Court of Justice in the case SAS Institute Inc. v World Programming Ltd:
“… the source code and the object code of a computer program are forms of expression thereof which ….. are entitled to be protected by copyright as computer programs, by virtue of Article 1(2) of Directive 91/250”
The most important exclusive right granted by copyright is in the name itself: the right to produce copies. Only the copyright holder is allowed to reproduce a work and to give new copies of that work to third parties. Additionally, only the copyright holder has the exclusive right to modify the work and to make it publicly available (for example, to offer it up online for download).
Do note that there is a distinction between a “copyright holder” and an “author”. The term “author” refers to the person who created the particular piece of copyrighted work. Meanwhile, the “copyright holder” is the person who has the exclusive rights over that work. Often, both the author and the copyright holder are the same person, as copyright automatically vests in the author from the instant that the work is created.
In most legal systems, copyright can be passed on to someone else, which is why the author and the copyright holder of a particular piece of work can, in some circumstances, be two different entities.
The Threshold of Originality
While copyright automatically vests in the author at the moment of the creation of a particular piece of work, this can only happen when that piece of work clears what is referred to as the “threshold of originality”. In other words, this means that the piece of work has to be sufficiently original to justify it being protected by copyright.
In the legal context, being “sufficiently original” here first indicates that the work must be independently created; in other words, the work cannot be copied or comprised of another pre-existing work. To this end, the Court of Justice of the European Union (CJEU) has clarified the status of the threshold of originality required for work to be copyrightable under EU law. In Infopaq International A/S v Danske Dagblades Forening, the CJEU affirmed that “copyright … is liable to apply only in relation to a subject-matter which is original in the sense that it is its author’s own intellectual creation.”
The EU Computer Programs Directive codifies these concepts specifically for source code. Under Article 1(3), this Directive states that:
“a computer program shall be protected if it is original in the sense that it is the author’s own intellectual creation. No other criteria shall be applied to determine its eligibility for protection.”
Nevertheless, it is also important to note that an “original” work for the purposes of copyright does not mean that the work has to be “unique”, or “novel”. Originality for the purposes of copyright therefore relates entirely to the origin or authorship of the work in question; it only needs to be original in relation to the author. Accordingly, a new work can be similar to others, but can still clear the threshold of originality to attract copyright protection.
It is entirely possible that two authors, without ever having met each other or discussed their work with each other, may create identical pieces of work. In such a situation, both these works would still each meet the threshold of originality, and therefore are subject to copyright protection, even if to an outside observer, they are not unique.
This emphasis on originality and not novelty was also further clarified by the CJEU, which stated that originality in EU copyright law is dependant on the “free and creative choices” of the author, signalling that the process by which the work was accomplished plays a role in determining whether the work is original enough to be protected by copyright, and not just the end result.
Clearing the Threshold
Because the uniqueness or novelty of the work is not a factor, the threshold of originality can be said to be a very low bar to clear. This can be seen in the UK case of THJ v Sheridan (while the UK is no longer part of the EU, UK copyright law was nevertheless interpreted in accordance with the relevant EU law in this case). This case concerned a software developer (“Mitchell”) who had developed a software program to assist with trading options. The defendant (“Sheridan”) would use Mitchell’s program to display data and information to his clients.
The copyright dispute in this case partially concerned the copyright in the images and charts generated by Mitchell with the help of his software. Mitchell claimed that these images and charts were eligible for copyright protection, while Sheridan claimed that, as what amounted to charts presenting data and information, the images were not sufficiently original to clear the threshold.
Example image produced using Mitchell’s software
The UK Court of Appeal, on observing such images, noted that:
- thought and care had to be put into how to lay out the information and other component parts of each chart;
- choices had to be made as to the fonts, colours, and other stylistic elements constituting each chart; and
- while each chart itself was not visually interesting, that in itself did not mean an absence of creativity.
Because of these factors, we see that even a non-unique, relatively uncreative work can still clear the threshold of originality for copyright protection, as long as it is:
- the author’s own intellectual creation; and
- made with free and creative choices.
For the context of software, that would mean that most developers’ original work would clear the threshold of originality and warrant copyright protection, that is if they wrote lines of source code by themselves without reference to other sources.
Being granted copyright protection is an important legal foundation for creating Free Software. This is because copyright protection gives developers the exclusive rights over their work. When developers then exercise choices in what kind of freedoms users of their work are able to enjoy (by licensing their work under a Free Software license, for example), copyright law functions to legitimize this choice to grant these rights to users. In other words, copyright gives developers a legal framework under which to license their original work under a Free Software license.
Works That Do Not Clear The Threshold
Nevertheless, there are always cases where some works may not clear this threshold of originality. Short phrases, data, simple symbols, or functional aspects of a particular piece of work are rarely considered copyrightable. For example, short phrases such as “World’s Greatest Dad”, or “better late than never” are not considered copyrightable. Similarly, lists of data or works compiled from public domain materials are also not considered to be copyrightable. The common theme in these uncopyrightable works is that they are too generic or simple to be considered an intellectual creation of the author.
Looking at this from a software perspective, a classic demonstration of non-copyrightable work would be the following program:
Here, both the function expressed and the phrase “Hello, world!” are simply too generic to be considered original works.
Developers will find that there are many such files or contributions in their own project repositories, such as files automatically generated by code, or config files that contain no creative expression. Similarly, very minor contributions like fixing a syntax error or correcting a spelling mistake can also be said to be non-copyrightable.
The Threshold of Originality and AI
With the rapid advancement of machine learning based generative systems (“AI”), questions have been raised about whether code generated by AI can be protected by copyright. These questions have not been definitively answered, as there currently lacks a uniform legal approach to classifying AI generated works.
This uncertainty has arisen because there is an unresolved debate about whether or not an AI generative system can be considered an author, thereby allowing AI generated works to be considered an intellectual creation made with free and creative choices of the author. While some consider that only human created works should be able to clear the threshold of originality, there are nevertheless other voices asking for copyright or similar legal protections over their investments in AI and AI generated products. Until such debates are definitively resolved in the law, copyright uncertainty will persist over AI generated works.
How To Deal With Uncopyrightable Files In Your Repository
The FSFE recommends the REUSE specification for developers to properly display their licensing and copyright information in their repositories. Generally, this would entail including a comment header in every file indicating such information. As the REUSE specification does not allow excluding files, some information still has to be recorded for the uncopyrightable files in your repo.
To deal with these files, you have two options:
- Simply use your regular copyright and license for this file. There is nothing that stops you from claiming copyright over your own works, even if a court might hypothetically find such files uncopyrightable.
- Waive your copyright using the CC0-1.0 license or another similar public domain dedication.
You may choose to use a copyright tag such as SPDX-FileCopyrightText: NONE to assert that there is no copyright holder.
Additionally, you may run into a situation where you are incorporating code that you think is not copyrightable into your software project. This could be because the code files are in your opinion simply not original enough to warrant copyright protection, or because they have been tagged in their original repos as having no copyright. In such situations, it helps for your project to implement and follow a strict compliance policy for code ingestion that includes processes to confirm the non-copyrightability of such code, including policies for scanning, tagging, and processing such code.
To that end, tools such as ScanCode, Fossology, and FOSSA exist to scan source code for identifying, among other things, copyright ownership of the code. The Linux Foundation has also made recommendations for developing such a compliance policy.
Why REUSE Recommends License Tags Even For Uncopyrightable Files
The reason that these files must be licensed is because opinions can differ about the threshold of originality. You may consider a file so insignificant that it does not merit licensing information, but downstream users may disagree and believe that the file in question falls under copyright, effectively disallowing them from using it if there is no license.
Such disagreements over copyrightability have even been litigated in European courts. While court cases have yet to provide an example directly relating to software and source code, they have produced examples where different European courts have had different opinions over whether the same piece of original work has cleared the threshold of originality.
A legal dispute between two Italian cosmetic brands demonstrates this difference clearly. The cosmetics brand KIKO S.p.A. (“Kiko”) claimed that a competitor WYCON Cosmetics (“Wycon”) had copied the design of its cosmetic stores, which included what Kiko claimed was a unique store layout. As a result, Kiko brought copyright infringement lawsuits against Wycon in a number of EU member states in which they both operated, including Italy, Belgium, and Portugal.
In Italy, the Italian Supreme Court of Cassation confirmed that Kiko’s store design reflected an expression of originality that warranted copyright protection, and that the similarities of Wycon’s shops amounted to copyright infringement. This was however not the opinion of the Belgian and Portuguese courts, which took the opposing view to the Italian courts. They were of the opinion that Kiko’s store layout and their arrangement of store components were standard features of cosmetics stores, and the store layout was essentially commonplace and generic, and did not require intellectual creativity to create. Kiko’s store design therefore could not pass the threshold of originality in their opinion.
In this example, we can therefore see a situation where a piece of work is able to be protected by copyright in one EU country, while failing to do so in another. This difference in how domestic law in different states may view the threshold of originality, even within the EU, is a good reason for developers to think about how to display copyright and licensing information in insignificant and uncopyrightable files.
Summary
In summary, the threshold of originality is a barrier to clear in order for a piece of work to be protected by copyright. Works that are of an author’s own intellectual creation and made with their free and creative choices will in the vast majority of cases clear that bar. That being said, there will be certain types of minor works in software, such as insignificant contributions, insignificant files such as config files, or generic lines of code, that taken by themselves will not warrant copyright protection.
As a legal concept, the threshold of originality is useful for Free Software developers to understand, so that they are able to discern which parts of their work can and cannot be protected by copyright. This in turn allows them to understand how to apply various standards and specifications, such as the FSFE’s REUSE specification, in a manner that helps to build a more informative and healthier Free Software legal ecosystem.
That being said, AI advancement has disrupted conventional understandings of originality and authorship. The discussion around whether or not AI generated works pass the threshold of originality is a complex one, which we will explore in a separate and future article.
Thanks to the amazing work of our volunteers, Ada & Zangemann can
now be translated on Weblate as well. This is a milestone for the
automation that is continuously being improved.
Some member of the FSFE team during LLW25