This post is also available in Deutsch, Français and Nederlands.
A recent decision from the Higher Regional Court of Cologne in Germany marked important progress for Cloudflare and the Internet in pushing back against misguided attempts to address online copyright infringement through the DNS system. In early November, the Court in Universal v. Cloudflare issued its decision rejecting a request to require public DNS resolvers like Cloudflare’s 1.1.1.1. to block websites based on allegations of online copyright infringement. That’s a position we’ve long advocated, because blocking through public resolvers is ineffective and disproportionate, and it does not allow for much-needed transparency as to what is blocked and why.
What is a DNS resolver?
To see why the Universal decision matters, it’s important to understand what a public DNS resolver is, and why it’s not a good place to try to moderate content on the Internet.
The DNS system translates website names to IP addresses, so that Internet requests can be routed to the correct location. At a high-level, the DNS system consists of two parts. On one side sit a series of nameservers (Root, TLD, and Authoritative) that together store information mapping domain names to IP addresses; on the other side sit DNS resolvers (also called recursive resolvers), which query the nameservers to answer where a particular website is located. The nameservers are like the telephone book listing names and phone numbers, while recursive resolvers are like the phone operator looking up a number.
While authoritative nameservers are managed and used directly by website operators, recursive resolvers are selected and used by those browsing the Internet. If you’re reading this at work, you may have navigated to this webpage using a DNS resolver chosen by your employer. If you’re reading it on a personal device at home, it’s possible you used your ISP’s default resolver. Alternatively, with a little technical know-how, you might have built your own DNS resolver and run it yourself or you might have chosen to use one of many public DNS resolvers available on the Internet.
Cloudflare launched its public DNS resolver, 1.1.1.1 in April 2018, because we wanted to provide a fast and private way to navigate the Internet. While Cloudflare’s resolver regularly scores as the fastest around, it is one of a number of options. Other well known public resolvers include Google’s 8.8.8.8, Cisco’s OpenDNS, and Quad9. Users might choose a public DNS resolver for privacy reasons, for added safety or security, or simply because they want the best performing option available. Whatever their reason, individuals can switch their DNS resolver at any time.
What does it mean to block through a DNS resolver?
Like other links in the Internet connection chain, DNS resolvers have sometimes been used as a way to try to prevent access to content. Blocking at the resolver level is like removing a listing from a phone book. By refusing to return an IP address in response to requests for a particular website, a DNS resolver can make it appear like an entire website has effectively disappeared from the Internet to an individual using that resolver. Unlike removing the content at the hosting provider, however, the content is still accessible online, just a bit harder to find. Much as having an unlisted phone number didn’t prevent a phone number from being found through other channels and called, a block in a resolver doesn’t preclude an Internet user from navigating to a website in a myriad of other ways. A user can use an alternative resolver, build their own resolver, or simply type in the website’s IP address.
Because DNS returns IP addresses for entire domains, blocking through DNS resolvers can only be done at a domain-wide level; it is not possible to block specific pieces of content, individual webpages, or even subdomains without blocking the entire website. So a blocking order seeking to remove a copyrighted image through DNS blocking — especially for a website with many contributors or user-generated content — would result in blocking all content on the entire domain. That means that unless the entire website is a problem, applying a block through DNS is likely to block access to content that has not been identified by a court as infringing or otherwise problematic.
The way DNS blocking works — declining to return an IP address — also means there is no explanation provided to an individual as to why they were unable to access the website at issue. There is no notice or transparency. Although there have been proposals for protocols that would allow an error code to be returned in such cases, nothing has yet been implemented.
Distinguishing public and private resolvers
Internet Service Providers (ISPs) located in particular jurisdictions have sometimes instituted blocks through their DNS resolvers as one way to try to comply with orders that apply in that jurisdiction directing them to make certain websites inaccessible to their users. For example, a German ISP that serves only German users might have its DNS resolver refuse to return an IP address for a website when provided an order by a German court to block that entire site.
Rightsholders have recently sought to extend such blocking to public DNS resolvers. But public DNS resolvers aren’t the same as DNS resolvers operated by a local ISP. Public DNS resolvers typically operate the same way around the globe. That means that if a public resolver applied the block the way an ISP does, it would apply everywhere. So the German court ordering the block would be dictating what information is available to the resolver’s users in India, the United States, Argentina and every other country the resolver is used. Attempting to apply blocks in a more geographically targeted way based on the location of individual resolver users raises serious technical hurdles not faced by local ISPs, and it also raises privacy issues worth taking seriously.
Cloudflare built 1.1.1.1 to allow Internet users an option for DNS resolution that would be fast and wouldn’t collect their personal information. Many DNS operators have historically sold information about users based on the websites they have queried – 1.1.1.1 is designed to prevent such information from ever being collected. Blocking orders directed at public resolvers would require the collection of information about where the requests are coming from in order to limit these negative impacts while demonstrating compliance. That would be bad for personal privacy and bad for the Internet.
These core features of public resolvers present fundamental obstacles to using such resolvers to block content.
Why blocking through public resolvers is not the solution to online abuse
Consider what you would expect if a website you were trying to visit had been blocked due to legal order. First, you would expect that the blocked content is genuinely prohibited by law. You would not expect an entire website to be unavailable merely because some portion of the website violated copyright, and you also would not expect a website to be blocked to a visitor in one country by virtue of an order issued in an entirely different country on the other side of the world.
Second, you would expect to be told why the website is unavailable. Rather than a blank screen or no response, you would want a message explaining that the website has been ordered blocked, and identifying the legal authority for that action.
Finally, you would expect that whatever blocking mechanism was instituted is actually effective. We should not be changing fundamental ways about how the Internet operates if it will not even have the intended effect.
Blocking through public resolvers fails all of these requirements. As discussed above, it cannot be applied narrowly to particular content or particular geographies. Unlike ISP blocking that is necessarily limited to the geographic region in which the ISP operates, blocking through global public resolvers can only be implemented in a way that extends across borders to jurisdictions that might never have sought to block the same content. That is, unless we collect more personal information than we need to about the user.
It’s also not transparent. A user does not know that they have been blocked from seeing content by a court order. They only know that they cannot access the website.That makes it hard for the public to hold government officials accountable for errors or overblocking.
And it’s not even effective. Traditionally, website operators or hosting providers are ordered to remove infringing or illegal content, which is an effective way to make sure that information is no longer available. A DNS block works only as long as the individual continues to use the resolver, and the content remains available and will become accessible again as soon as they switch to another resolver, or build their own.
The court in Universal rejects DNS blocking
Despite these problems, some rightsholders have insisted that public resolvers can be ordered to block websites based on online infringement. Cloudflare, along with others like Quad9 and Google, have pushed back. While there have been a limited number of preliminary rulings on this issue, the Higher Regional Court’s decision in Universal marks the first time that an appellate court in Europe has ruled on public resolver blocking in the main proceedings.
Originally filed in 2019, the Universal case was one of the first attempts by a rightsholder to obtain an order requiring blocking through a public DNS resolver. The case concerns an allegedly copyright infringing music album posted on a website that, at the time the case was filed, was using Cloudflare’s pass-through security and CDN services. The Cologne Regional Court issued a preliminary ruling directing Cloudflare to block the website through both our CDN service and our public resolver. Cloudflare has no mechanism for blocking websites through 1.1.1.1., and we have never blocked a website through our public resolver. But Cloudflare did take steps to block access to the website in Germany through our CDN and pass-through security service. The website subsequently went offline and is no longer available on the Internet. Recognizing the importance of the underlying legal principles at stake, we nonetheless continued to litigate the case.
The Higher Regional Court’s recent decision makes clear that public DNS resolvers are not an appropriate tool for seeking to address online infringement, or moderate content more generally. The court explained that “with the DNS resolver, the defendant provides a tool that is accessible to everyone free of charge, is in the public interest and is approved, and which participates purely passively, automatically and neutrally in the connection of Internet domains.” It further noted that blocking through a public resolver is not effective, because individuals can easily change resolvers.
Importantly, the court held that DNS services are protected by the EU’s Digital Services Act (DSA), which was enacted last year. Like the e-Commerce Directive before it, the DSA recognizes that different types of services have different abilities to address content issues, and it distinguishes “mere conduit” and “caching” services from “hosting” services in their roles in addressing infringing content. Helpfully, the DSA expressly lists DNS and CDN services as non-hosting services subject to different obligations than hosting services. The Higher Regional Court recognized that DNS resolvers are entitled to the same protections from liability as other “mere conduits,” and it rejected the plaintiff's request for DNS blocking in this case.
The battle continues
While the Higher Regional Court’s decision represents important progress on the DNS issue, the fight over how best to address online infringement continues. Rightsholders have filed lawsuits against other DNS providers and in other jurisdictions seeking similar blocking orders. We will continue to advocate against that outcome, because we think it is bad for the Internet. We hope that the Higher Regional Court’s reasoning on the DNS issue will help persuade other courts.
At the same time, while the Universal decision on DNS is the headline, there were other parts of the opinion that raise concerns. The court affirmed the lower court judgment requiring Cloudflare to block access to the website at issue through our CDN and pass-through security service. That decision has no immediate practical effect, because the website at issue is no longer available online and Cloudflare was already in compliance with the judgment. But to the extent the decision can be read to imply a broader obligation by pass-through security and CDN services to address online content, that is inconsistent with the nature of our services and with the DSA, which expressly identifies CDN services as among the caching services entitled to a liability privilege. Cloudflare therefore plans to appeal that aspect of the decision.
We appreciate the efforts of thoughtful judges to learn about how the Internet works and make sure their decisions are consistent with the larger public benefits of a well-functioning Internet, including security, reliability, and privacy. This decision marks further progress in Cloudflare’s fight to ensure that efforts to address online infringement are compatible with the technical nature of various Internet services, and with important legal and human rights principles around due process, transparency, and proportionality. We will continue that battle both through public advocacy and, as necessary, through litigation, as one more part of helping build a better Internet.