on intermediary liability


1. Why is intermediary liability important?

Given the amount of potentially unlawful or harmful content that is transmitted through their services and their technical capabilities to control access to that content, internet intermediaries are under increasing pressure from governments and interest groups to act as ‘gatekeepers’ of the Internet. This is usually done through the adoption of laws that hold intermediaries financially or criminally responsible if the intermediary fails to filter, block or remove content which is deemed illegal. This often results in private companies censoring content on behalf of the state without appropriate safeguards or accountability mechanisms.


 2. What are the different models of intermediary liability?

There are several models of liability around the world. These can be broadly categorized into three models:

Expansive protections from liability – Intermediaries are not responsible for content produced by others so long as they have not intervened in that content. The intermediary is only required to restrict content when ordered to do so by a court or other independent adjudicatory body. For example, this is the model in the United States (except for alleged copyright infringements) and in Chile.

Conditional immunity from liability – Intermediaries are required to comply with predefined procedures to be granted immunity from liability of content on their platforms and networks. Two distinct systems of the conditional immunity or safe harbor model currently in practice include:

Notice and Takedown - The intermediary restricts content on receipt of a notice from a third party as per legal procedures

Notice and Notice - The intermediary passes on a third party notice to the user who uploaded the content.

For example in Singapore, Ghana, Uganda, South Africa and the EU, intermediaries are effectively immune from liability if they comply with notice and takedown procedures. Canada is an example of a conditional immunity model of liability that follows a notice and notice procedure.

Primary liability – Intermediaries can be held liable for all content on their platforms and networks and consequently, may face both criminal and civil sanctions. This often leads to intermediaries monitoring, identifying, and pro-actively removing content in order to avoid sanctions. For example, in Thailand intermediaries are generally held primarily liable for content and this leads them to actively monitor the content on their networks.


3. What model of intermediary liability do the Manila Principles support?

The Manila Principles support expansive protections from liability for intermediaries, derived from the notion that intermediaries should not be held liable for content produced by others and may only remove content when ordered to do so by a court. The fact that intermediaries have the technical means to prevent access to content does not qualify them as the best placed to evaluate the ‘illegality’ of the content in question. Such a determination should be primarily a matter for an independent judicial body, and not a private company.

However, the Manila Principles recognise that a notice-and-notice model may also be appropriate to deal with issues that do not involve allegations of serious criminality such as defamation or copyright claims. For this reason, the Manila Principles recommend a number of procedural safeguards that should be included as part of a safe harbour model with a notice-and-notice system in order to be compatible with human rights standards (see question 9).

While the Manila Principles reject notice-and-takedown procedures and the primary liability model, we stress that at the bare minimum, where such models and procedures exist, they must lay down procedural safeguards that must be respected in order to protect freedom of expression.


4. Do the Manila Principles support the use of a graduated response or “three strikes” system for copyright infringements?

No. Although it is permissible under the Manila Principles to forward a notice to a user that their content has been reported as allegedly infringing the law, this would not require an intermediary to forward notices that a user has merely accessed allegedly infringing content through an intermediary. On a related point, the Manila Principles also specify that intermediaries should not be required to disclose personally identifiable user information without a judicial order.


5. What should a content host do on receipt of a notice from a third party that asks for a user's content to be removed?

The first thing the host should consider is the format of the removal request—is the notice an order from a court with jurisdiction over the host? If not, the host should consider doing nothing. If the law does not make the host liable for that content, either immediately or after receipt of such a notice (and the Manila Principles recommends that it shouldn't), then the most the content host should do is to forward the notice to the user who uploaded the content. This allows users to act on alleged illegal content if they wish to, by either complying with the notice for removal or contesting it in the court. In some cases (particularly where large volumes of requests are made), it will be proper for the intermediary to charge for this on a cost recovery basis.


6. Do the Manila Principles also apply to search engines and domain providers?

Broadly, search engines and domain hosts are treated similarly to other intermediaries, in that, they should not restrict access to any content (which would include removing a search result or blocking a domain name) without a judicial order. An exception to the treatment of intermediaries who do not host content, is that they are not expected to pass on notices of alleged illegality (after all, they may not even know who the content provider is).


7. What if a third party requests content to be removed and that content is also against the host's terms of service?

In that case, the restriction of the content is within the discretion of the host, and not strictly a response to the notice at all. But the host should avoid allowing its terms of service to be used as a back door for streamlined content removal by third parties. How can it avoid this? By adopting and incorporating the Manila Principles on transparency, accountability and procedural fairness. These include making sure that its policies are clearly expressed ahead of time, giving the user the chance to appeal the decision to restrict the content pursuant to those policies, and disclosing the restriction in its transparency reporting.


8. But what if the law requires the content to be removed on receipt of a notice from a third party (“notice and takedown”)—does that make the Manila Principles irrelevant?

No, though an intermediary has to comply with the law, they can limit the impact of the content restriction by following other aspects of the principles; as explained in the next answer.


9. How should an intermediary minimize the impact of complying with an order to remove content on users rights?

For example, intermediaries should forward the notice to the user who uploaded the content and, in doing so, describe any available counter-notice or appeal mechanisms that the user can use to have the content reinstated. Intermediaries should also make sure that they have the technical means to reinstate the content, if such an appeal succeeds. And they should limit the restriction of the content, in terms of how much is restricted, over what geographical area, and for what time period, to the minimum necessary. If the service they provide allows them to display a notice in the place where the restricted content would otherwise be found, they should also do that.


10. But some content really does need to be taken offline without a court order—what do the Manila Principles recommend in such a situation?

Emergency circumstances do sometimes arise. In most of these cases, the intermediary may have discretion to act within their terms of service, as explained above. But there should still be no compulsion on an intermediary to act without a judicial order. The Manila Principles do, however, allow for such an order to be made without a judicial hearing in cases that are clearly and narrowly defined by law as exceptional circumstances (we do not attempt to itemize what these circumstances might be). Even in those cases, a hearing to review the order and the implementation of the order must subsequently be available.


11. Why don't the Manila Principles place limits on the grounds on which intermediaries can limit content that they host under their terms of service?

In general, intermediaries who host content ought to be able to decide what they do and don't want to host; this freedom has contributed to the development of the Internet into such a diverse and dynamic ecosystem.  For this reason we do not wish to compel intermediaries to accept user content for hosting, or seek to evaluate the reasons why they choose not to (just as the Principles do not evaluate the various substantive grounds on which content may be prohibited by law).  Instead, the Principles require that whatever terms of service intermediaries apply must be clear and transparent, and that a user must be able to appeal against their application.  Intermediaries are also obliged to respect human rights, and governments to ensure that they do so.  Finally, while the Manila Principles do not cover the substantive content of intermediaries' terms of service, there are various other projects that do.