Build what's next on GitHub, the place for anyone from anywhere to build anything.
Join us October 28-29 in San Francisco or online for GitHub Universe, our flagship developer event uniting people, agents, and the world's code.
The EU Copyright Directive has come down to a final vote. Read on for more on what’s at stake and how the developer community shaped the debate.
TL;DR – The highly controversial Copyright Directive proposal is headed to a final vote. While copyright reform is needed, there are problems with the specific approaches outlined in this proposal. Developers took part in discussions and raised concerns about unintended negative impacts on software development. That helped policymakers narrow the scope of possible impact on software and extended the profile of open source software in the policy arena. Read on for the full story.
Last September—two years after the EU Commission first proposed the Copyright Directive, we filled you in on the status of negotiations. At that time, the EU Parliament had just adopted its version of what the law should look like.
Remember, all three EU institutions—the Commission, Council, and Parliament—have to agree on a single version of the Copyright Directive in order for it to become law. The Parliament’s decision in September triggered joint negotiations among the Commission, Council, and Parliament (called “trilogues”) in October. The three bodies had tried since then to agree on a single version but reached an impasse in January. On February 8, 2019, the Council revised its position in the hopes of enabling a breakthrough in the trilogues. That’s why the EU Commission, Council, and Parliament met this week to discuss the differences in each of their versions of the Copyright Directive and try to come up with a single version together. This time, they were able to reach an agreement, and GitHub remains out of scope.
Next, this jointly approved version will go to both the Council and Parliament for final yes-no votes to decide whether it becomes law. At this stage, each body can either accept or reject the proposed law. They can also still pass the law while accepting or rejecting certain articles within it, but don’t expect to see any further changes to the wording of the articles. Open source software developing and sharing platforms like GitHub should remain out of scope.
Let’s take a look at the diff from our September 2018 blog post to see how our analysis of the Copyright Directive has changed since Parliament’s September vote:
-On September 12, the EU Parliament voted to: +On February 11, the EU Parliament, Council, and Commission voted +to: * Make content-sharing platforms directly liable for copyrighted content that users upload, which could lead to use of upload filters (Article 13) -* Exclude "open source software developing platforms" from that - liability and need for upload filters (Article 2) +* Exclude "open source software developing and sharing platforms" + from that liability and need for upload filters (Article 2) -* Allow an exception for text and data mining only by research - institutions for scientific purposes on a "non-for-profit" basis, - with only an "optional" exception for others (Article 3) +* Create a mandatory exception for text and data mining on large + datasets but with an opt-out for rightholders who do not want + the exception to apply to their works (Article 3) -* Create a new right for press publishers to require a license to - use content of news articles except for "mere hyperlinks, which - are accompanied by individual words" (Article 11) +* Create a new right for press publishers to require a license to + use content of news articles except for “individual words” or “very + short extracts” (Article 11) +* Exclude software developers from articles related to the + principle of appropriate and proportionate remuneration. (Those + articles cover reporting requirements and contract adjustment + rights related to remuneration for rightholders, and resolution of + disputes about reporting or contract adjustment rights. They also + provide the right to revoke an exclusive license or transfer of rights + where the copyrighted work hasn't been exploited.)
-On September 12, the EU Parliament voted to:
+On February 11, the EU Parliament, Council, and Commission voted
+to:
* Make content-sharing platforms directly liable for copyrighted
content that users upload, which could lead to use of upload
filters (Article 13)
-* Exclude "open source software developing platforms" from that
- liability and need for upload filters (Article 2)
+* Exclude "open source software developing and sharing platforms"
+ from that liability and need for upload filters (Article 2)
-* Allow an exception for text and data mining only by research
- institutions for scientific purposes on a "non-for-profit" basis,
- with only an "optional" exception for others (Article 3)
+* Create a mandatory exception for text and data mining on large
+ datasets but with an opt-out for rightholders who do not want
+ the exception to apply to their works (Article 3)
-* Create a new right for press publishers to require a license to
- use content of news articles except for "mere hyperlinks, which
- are accompanied by individual words" (Article 11)
+* Create a new right for press publishers to require a license to
+ use content of news articles except for “individual words” or “very
+ short extracts” (Article 11)
+* Exclude software developers from articles related to the
+ principle of appropriate and proportionate remuneration. (Those
+ articles cover reporting requirements and contract adjustment
+ rights related to remuneration for rightholders, and resolution of
+ disputes about reporting or contract adjustment rights. They also
+ provide the right to revoke an exclusive license or transfer of rights
+ where the copyrighted work hasn't been exploited.)
To sum up, Article 13 would still impose liability on platforms that will lead to the use of upload filters. In the earlier version, “open source software developing platforms” were excluded from scope. Now that exception is a little broader: “open source software developing and sharing platforms.” The optional exception for text and data mining became mandatory. On the new right for press publishers, the earlier version said “mere individual words” and now it’s “individual words or very short extracts.” One completely new addition since Parliament’s vote in September is that software developers are excluded from a set of remuneration-related articles.
Now let’s see how the implications for software have changed by comparing the Copyright Directive’s latest version to what we originally wrote in our September blog post:
- If Parliament's version of the Copyright Directive becomes - the law: + If the Parliament and Council each vote to approve the latest + version of the Copyright Directive so it becomes the law: -* Sites that host user-generated content may need to filter - content that users upload, but "open source software developing - platforms" (like GitHub) wouldn't need to. We supported a broader - exclusion for software development platforms, archives, and - repositories, as it would have protected more of the software - development community. However, Parliament adopted the - narrower language. Since elements of software development - happen beyond that narrow exclusion, developers would need to - consider whether they might be subject to liability for the content - they host and resort to measures like filtering. +* Sites that host user-generated content may need to filter content + that users upload, but "open source software developing and + sharing platforms" (like GitHub) wouldn't need to. We supported + a broader exclusion for software development platforms, archives, + and repositories, as it would have protected more of the software + development community. However, the three EU institutions + adopted narrower language though they broadened it slightly by + adding “and sharing” platforms. Since elements of software development + happen beyond that narrow exclusion, developers would need to consider + whether they might be subject to liability for the content they host + and resort to measures like filtering. -* Developers may need licenses to mine content—including for - artificial intelligence and machine learning—unless individual - EU countries decide to adopt an exception from text and data - mining that would cover them. Without a mandatory broader - exception, developers would be subject to a patchwork of regulations - across different EU countries. +* Individual EU countries decided to make the broader exception from + text and data mining, which could apply to mining content for artificial + intelligence, mandatory. Without a mandatory broader exception, + developers would have been subject to a patchwork of regulations + across different EU countries. -* Developers who link to news articles may need to pay to use - content like article headlines or snippets. It may take a judge to - interpret what the phrase "individual words" means exactly in the - hyperlinks press exception we called out above. In the meantime, - developers would need to be careful about what content they - include to describe links. +* Developers who link to news articles may need to pay to use + content like article headlines or snippets. It may take a judge to + interpret what the phrase "individual words" or “very short extracts” + means exactly in the hyperlinks press exception we called out above. + In the meantime, developers would need to be careful about what + content they include to describe links. +* Software development platforms aren't subject to remuneration- + related reporting and contract requirements that are irrelevant to + open source software developers.
- If Parliament's version of the Copyright Directive becomes
- the law:
+ If the Parliament and Council each vote to approve the latest
+ version of the Copyright Directive so it becomes the law:
-* Sites that host user-generated content may need to filter
- content that users upload, but "open source software developing
- platforms" (like GitHub) wouldn't need to. We supported a broader
- exclusion for software development platforms, archives, and
- repositories, as it would have protected more of the software
- development community. However, Parliament adopted the
- narrower language. Since elements of software development
- happen beyond that narrow exclusion, developers would need to
- consider whether they might be subject to liability for the content
- they host and resort to measures like filtering.
+* Sites that host user-generated content may need to filter content
+ that users upload, but "open source software developing and
+ sharing platforms" (like GitHub) wouldn't need to. We supported
+ a broader exclusion for software development platforms, archives,
+ and repositories, as it would have protected more of the software
+ development community. However, the three EU institutions
+ adopted narrower language though they broadened it slightly by
+ adding “and sharing” platforms. Since elements of software development
+ happen beyond that narrow exclusion, developers would need to consider
+ whether they might be subject to liability for the content they host
+ and resort to measures like filtering.
-* Developers may need licenses to mine content—including for
- artificial intelligence and machine learning—unless individual
- EU countries decide to adopt an exception from text and data
- mining that would cover them. Without a mandatory broader
- exception, developers would be subject to a patchwork of regulations
- across different EU countries.
+* Individual EU countries decided to make the broader exception from
+ text and data mining, which could apply to mining content for artificial
+ intelligence, mandatory. Without a mandatory broader exception,
+ developers would have been subject to a patchwork of regulations
+ across different EU countries.
-* Developers who link to news articles may need to pay to use
- content like article headlines or snippets. It may take a judge to
- interpret what the phrase "individual words" means exactly in the
- hyperlinks press exception we called out above. In the meantime,
- developers would need to be careful about what content they
- include to describe links.
+* Developers who link to news articles may need to pay to use
+ content like article headlines or snippets. It may take a judge to
+ interpret what the phrase "individual words" or “very short extracts”
+ means exactly in the hyperlinks press exception we called out above.
+ In the meantime, developers would need to be careful about what
+ content they include to describe links.
+* Software development platforms aren't subject to remuneration-
+ related reporting and contract requirements that are irrelevant to
+ open source software developers.
These changes mean that developers might be subject to liability for the content they host and might have to resort to measures like filtering if they’re hosting content outside of “open source software developing and sharing platforms” (again, the phrase “and sharing” is new and adds to what’s excluded). The broader text-and-data-mining (TDM) exception narrows the circumstances in which developers need licenses to mine content on large datasets, which is critical to the development of artificial intelligence and machine learning in Europe. The fact that this TDM exception is mandatory means that each EU country has to make it part of its national law, though the article also allows publishers to opt out of allowing this exception to apply to their works. On Article 11’s new right for press publishers, it’s still unknown how courts will interpret what counts as “individual words or very short extracts” to be excluded from scope.
On a bigger-picture level, copyright reform is well overdue. Some elements of the Copyright Directive, like protecting the public domain and facilitating access to out-of-commerce works, are a positive step toward updating the legal framework to better reflect today’s digital world.
But Articles 11 and 13 would undermine protections that have made the internet thrive as a place to collaborate creatively and access knowledge—for all kinds of content, not just software code. We talked about implications for developers earlier in this section, but people who use the internet to write, research, or work together to build something in countless other contexts stand to lose when snippets of texts come with a price tag and when uploaded content might be unfairly censored. So it’s not surprising that these two articles have been so controversial throughout the negotiations. Check out Communia’s flow chart for a visual interpretation and explanation of what Article 13 would do.
Over the past many months, you’ve joined us in helping policymakers understand how iterations of the Copyright Directive’s text could impact software. We focused mainly on Article 13 because it proposed a requirement on certain information-sharing services to filter uploaded content for potential copyright infringement, and, initially, that category would have included software development platforms. Filters aren’t likely to be effective in detecting infringement in software code, given the number of contributors, licensing of dependencies, and complexity of licensing requirements. And false positives would cause dependencies to disappear, breaking software builds.
So you spoke up—and policymakers listened. It took some follow-up to get the nuances of the language right, but ultimately, Article 13 excludes “open source software development and sharing platforms” from its scope.
That means that even if the directive passes, platforms like GitHub won’t have to filter or take other measures to monitor our users’ uploads. Just speaking for software, we can say that we’ve protected the vibrant, critical online collaboration that supports the apps, websites, programs, and everything else that developers build. We’ve also protected the economies and services that rely on them—all thanks to you.
Beyond the concrete improvements to specific pieces of the Copyright Directive, developers increased visibility of open source, and software development overall, to policymakers. After we first raised concerns about the Copyright Directive, policymakers told us they weren’t thinking about software development when they drafted this proposal. All three EU institutions involved in the negotiations later recognized that software was inadvertently in scope because code is protected under copyright law, and that they needed to alter the proposal to protect software. So, they did. They even went so far as to add a new article, “Exclusion of software developers” related to a section on remuneration, recognizing that those rules shouldn’t apply to open source software developers since open source licenses don’t provide for remuneration for rightholders. Open source sustainability is a separate topic we’re working on but suffice it to say, restricting access to code through copyright law would undermine the openness that the open source ecosystem relies on.
Above all, you energized collaboration within the open source community (from developers, to researchers, to businesses) and across country borders, which sets the stage for more collective action to protect software. It’s still an open question whether the Copyright Directive will ultimately be approved, but whatever happens, your collaboration has made policymakers realize that they need to consider software development and the open source software ecosystem in their decision-making process. You’ve left a lasting impact on how software and open source are viewed not just in the policy space, but around the world.
The Council and Parliament will decide whether to approve or reject the law in March or April. Contact your Council members and Members of Parliament to thank them for protecting software development, and to explain what issues remain for the broader internet in Articles 11 and 13.