Overview of Types of Content Takedowns and Reporting

Below is an overview of the six most common types of content takedowns that internet and telecommunications companies can and have been reporting on. The first five are types of content takedowns or blocking done in response to certain types of legal requests or demands; the sixth category regards content takedowns based on a company’s own content policies.

This category focuses on government-issued legal demands, as well as other legal demands for content takedowns issued by individuals asserting a violation of local law (e.g., defamation). Although copyright and trademark requests are also forms of legal demands, companies typically report on them separately. Therefore we also separately evaluate company performance on reporting about copyright and trademark requests, later in our survey.

Currently, most companies lump together government and non-government legal demands into one report, rather than counting them separately, and as a result our evaluation of their reporting in Charts 2a and 2b also assesses them collectively. However, a few companies, such as Daum Kakao and Apple, report separately on at least some forms of government and non-government requests. Going forward it is recommended that all companies do so, for greater clarity and transparency.

Reporting on government and other legal demands is currently the most prevalent form of content takedown reporting across both the internet and telecommunications sectors. Reporting on this category of demands generally includes data on the number of content demands a company received, how many accounts or items these demands impacted and how the company responded to such demands.

When discussing impact, it is important to note that some companies do not always respond to demands by completely removing content from their entire platforms or networks. Rather, they may otherwise restrict or geo-block content, like Facebook does, or block access to certain IPs and URLs in a country or region if the applicable law is regionally specific, like Telefonica does.

As a result, some internet platforms also report specifically on content restrictions short of complete takedown, and some telecommunications companies also report on IP/URL or content blocking.

One of the most common types of legal requests seeking content takedown are those alleging copyright infringement. Among the companies we surveyed, the most commonly cited copyright law is the Digital Millennium Copyright Act (DMCA), a U.S. statute that enables copyright holders to file requests or notices for content takedown based on the premise that a certain piece of posted content infringes upon copyright. Most internet intermediaries that are based in the United States or that largely operate in the United States follow the DMCA’s guidelines on copyright. However, international internet intermediaries whose primary hubs of operation are outside the United States tend to receive content requests based on their own local copyright protections and laws.

Reporting on copyright takedowns is currently the second most prevalent form of reporting on content takedowns. Reporting on this category of demands typically includes data on the number of related requests a company received and how they responded. However, this data often excludes vital points such as information on the number of counter-notices a company has received and how much content was restored as a result.

Just as copyright holders issue requests demanding the takedown of allegedly copyright-infringing content, so do trademark holders who submit takedown requests alleging infringement of their marks. Among the companies we surveyed, the majority of internet platforms evaluate trademark requests based on U.S. trademark law. However, international internet intermediaries whose primary hubs of operation are outside the United States receive content requests based on their local trademark protections and laws.

Reporting on trademark takedowns typically includes data on the number of requests a company received, how a company responded to those requests and the number of accounts or items that were impacted by those requests. Compared to other forms of reporting on intellectual property-related takedowns such as copyright, reporting on trademark takedowns is not as prevalent on communications-centered websites such as social media platforms, although it is more prevalent on e-commerce platforms. Reporting on trademark takedowns is, however, gradually becoming a more widespread practice across internet platforms that host user-generated content such as blogs.

This category includes government-mandated network shutdowns of internet services as well as incidents in which telecommunications systems, online platforms, or other internet-based services have been disrupted by government agencies. This category does not include demands to block access to specific pieces of content, IP addresses, or URLs, but rather focuses on system-and service-wide shutdowns and interruptions.

Network shutdowns primarily impact telecommunications operators. However, internet disruptions may also be targeted at companies that offer services over the internet in addition to or instead of the telecoms themselves, especially during times of political turmoil and instability. For example, a government may force a shutdown of a telecom’s internet service, or it may force that telecom to disrupt access to (e.g.) Twitter alone. Both shutdowns and disruptions aim to bar access to certain services or the internet in general.

Although network shutdowns are becoming increasingly prevalent across the world, internet and telecommunications platforms do not typically document or report on instances of shutdowns or disruptions in their transparency reports. Sometimes this is due to legal restrictions which prevent such disclosures in countries that companies operate in.
Companies that report on this category well typically disclose data on the location, timing, and reason behind the shutdown. A handful of companies also provide quantitative data on the number of shutdowns and the duration of these shutdowns.

In May 2014, the Court of Justice of the European Union ruled that citizens of the EU could request search engines to delist search results tied to their names if the information in the result was “inadequate, irrelevant or excessive in relation to the purpose of processing.” This right to erasure was also included in the EU’s General Data Protection Regulation which came into effect in May 2018. A similar Right to Erasure law came into effect in Russia in January 2016.

Although reporting on delisting requests received as a result of the Right to be Forgotten laws are a form of company reporting on legal requests, companies have begun reporting on these requests separately. As a result, we have assigned this form of reporting its own category.

Currently, transparency reporting around Right to be Forgotten delisting requests is the least prevalent form of content takedown-related transparency reporting—only Microsoft and Google currently issue such reports (for Bing and Google Web Search, respectively). This is most likely because these regulations are directed specifically at search engines, and as a result other types of services don’t receive—or at least, shouldn’t be receiving—these requests. However, to the extent services outside the scope of these laws are receiving such requests, that is something the public and policymakers should know, and the best practice for those companies would be to report on those requests—especially because the Right to be Forgotten may ultimately be expanded to cover them.

A company’s Community Guidelines (also often referred to as Community Standards or Terms of Service) outline its self-defined standards and norms for what type of content is acceptable and unacceptable on its service. Companies often remove or restrict content, or suspend or deactivate accounts, based on violations of such guidelines, whether that violation was flagged by users or discovered by the service through its own automated or human review. The need for transparency and accountability in this area is particularly strong because the company is essentially enforcing its own private law, rather than complying with public law subject to the typical checks and balances of the political process.

Reporting on Community Guidelines-based content takedowns is still a new and emerging practice. Until recently, the very few platforms that reported on Community Guidelines-based content takedowns at all only did so for one or another category of content (for example, Microsoft only reports on nonconsensual pornography takedowns, while Automattic only reports on terror content takedowns). However, over the past few months major internet platforms such as Google (via YouTube) and Facebook have issued more comprehensive transparency reports covering Community Guidelines-based takedowns across multiple content categories.

Each of these six types of reporting on content takedowns are integral to understanding the nature of and limitations on online freedom of expression today, and we recommend that companies issue reports providing data on all of the above content demands where relevant to their services and the types of requests they receive. However, there is currently a significant amount of variance between how and what internet and telecommunications companies report on.

Charts 1a and 1b summarize which of the six types of content demands the 35 companies we surveyed are (or are not!) reporting on. Two important notes for understanding these charts: First, note that for the Community Guidelines-based content takedowns section, a company earned a check mark if it published any data on its Community Guidelines-based content takedown efforts. However, the quality and extent of the reported data varied widely, and as previously noted only two companies are currently offering anything close to comprehensive numbers. Second, these charts reflects the first of our 11 general best practices—regular issuance of a report over consistent periods—by highlighting each company’s last reported period. Therefore that best practice is not addressed in later charts that assess compliance with the overall best practices.