Protection Gaps in Public Law Governing Cyberspace: Israel’s High Court’s Decision on Government-Initiated Takedown Requests

Introduction

On April 12, the Israeli High Court of Justice issued a decision rejecting a challenge by two Israeli nongovernmental organizations—Adalah and the Association for Civil Rights in Israel—to the legality of the practice by the Cyber Unit in the State Attorney’s Office of requesting online platforms to take down unlawful content. The court held, however, that the practice in question, which is both prevalent and regulated by internal procedures, is a government act that must be based on a general or specific authorizing statute. Furthermore, it held that government-initiated takedown requests might infringe on basic constitutional rights, including the freedom of expression, and that, were such infringements proved, the practice would have been deemed unlawful without an explicit legislative mandate. Still, in the particular circumstances of the case, the court found no evidence of infringement; hence, it held that the practice could continue to be based on the general residual power of the government, as established in Basic Law: The Government, Article 32, to take lawful action in the name of the state wherever no other authority has been charged with undertaking the same action.

The Petitions

Israeli law contains a number of specific legal provisions authorizing judges to order online platforms to take down illegal content in cases involving online gambling, pedophilia, prostitution, drug trafficking, acts of terror organizations and disclosure of sensitive private information. There is no statutory basis for requesting or ordering online platforms to take down other illegal content, including hate speech, incitement to violence, sexual harassment and bullying. A draft bill proposing to authorize courts to issue orders in such cases (colloquially referred to in Israel as the Facebook bill) was strongly resisted by the major online platforms and was eventually blocked in 2018 at the request of Prime Minister Benjamin Netanyahu.

In the absence of a comprehensive statutory basis for the removal of harmful online content, the Cyber Unit of the State Attorney’s Office developed an alternative enforcement track, which it dubbed the “voluntary” track. Under this track, the Cyber Unit refers takedown requests pertaining to what it considers to be illegal content to the major online platforms. The content in question is typically brought to the attention of the Cyber Unit by Israeli security agencies. The Cyber Unit then conducts an internal process for reviewing the legality of the content in question and the propriety of seeking its removal. In practice, about 90 percent of the requests made by the Cyber Unit to the platforms are accepted. Most of the requests pertain to terrorism offenses and incitement to violence (99 percent of the requests made in 2018 fall under this category); other requests pertain to cases of intimidation, privacy breaches or harassment of public officials. By far, most of the requests to date were directed at Facebook (87 percent); Twitter is a distant second (8 percent), and the remaining requests were directed at other online platforms (YouTube, Instagram, Google and others).

In Adalah v. Cyber Unit, the two petitioners claimed that the operations of the Cyber Unit infringed the constitutional right to freedom of expression and the right to due process (both extrapolated by the court in previous cases by way of interpretation of Basic Law: Human Dignity and Liberty). Petitioners also claimed the Cyber Unit operated without adequate statutory authorization. In particular, they maintained that the Cyber Unit violated basic principles of constitutional law in that it unilaterally determined the scope of freedom of expression and whether a criminal offense was committed, without referring the matter to a court of law nor allowing those affected by their decision to challenge it. It cannot be ignored, according to the petitioners, that political content covered by the constitutional right to freedom of expression was subject to removal requests by the Cyber Unit.

The High Court’s Decision

The first major issue addressed by the court in its decision is the characterization of the voluntary track. The Cyber Unit maintained that its activities do not entail the exercise of any government authority, as the decision whether or not to remove content remains exclusively with the online platforms. Arguably, the Cyber Unit requests takedowns in lieu of private individuals who lack the expertise and know-how to communicate effectively with the online platforms. The court rejected these claims and held that the Cyber Unit is clearly engaging in a government act, which requires some form of statutory authorization (general or specific). This is because the Cyber Unit has the power to subject the online platforms to mandatory regulations should they systematically refuse to comply with its takedown requests. The court further held that the practice of the Cyber Unit has the potential to infringe basic constitutional rights, and it noted that if such infringement were shown to it, then specific statutory language authorizing the practice would have been required.

However, the court found that the petitioners failed to establish a connection between the practice of the Cyber Unit and the infringement of any specific constitutional rights. The arguments invoked by the court in support of this conclusion are hardly persuasive, and one may speculate that the court went out of its way to not prohibit the voluntary track, as it seemed to consider this track an effective stopgap measure to address—pending the passage of some version of the Facebook bill—the many risks associated with the current online environment. The court’s decision can therefore be regarded as an attempt to signal to the parties that the practices of the Cyber Unit raise serious public law problems but that, because of the petition’s specific procedural and evidentiary shortcomings, the time has not yet come to declare such practices to be unlawful.

The first flaw the court found in the petitioners’ case was their failure to provide evidence of takedown requests resulting in the infringement of a constitutional right, such as the freedom of expression. However, evidentiary harm is precisely one of the harms stemming from an enforcement modality that relies on voluntary and nontransparent cooperation between governments and online platforms: Affected individuals are aware that content they posted was removed by an online platform because of incompatibility with the applicable community standards or terms of use; they are not aware of the fact that the platform acted in response to a government takedown request. Nor are they typically aware of the existence of a direct and discreet line of communication between the government and the platforms. Under these structural conditions of an “invisible handshake,” it is not clear how the court expected the petitioners to present evidence of specific infringements caused by government requests.

A second point developed by the court in support of its conclusion that no infringement of constitutional rights was shown is that much of the removed content was presumably attributable to bots and avatars, which, according to the court, meant there was no human subject behind the content but, rather, computerized objects, artificial intelligence, or a network of hostile users, none of which possess human rights. This is also a controversial statement, as the question of whether virtual entities enjoy digital rights (in a manner comparable to corporations that enjoy rights under law) is far from settled in legal theory and comparative law practice (see, for example, Charter of Human Rights and Principles for the Internet, Article 8(d)). Still, even if this is the case as a matter of lex lata, its relevance to the case at hand is dubious since, at present, artificial intelligence is hardly responsible for generating new social media content (as opposed to disseminating content or interacting with human-generated content). The real question in cases involving bots and avatars is whether or not to protect this form of anonymized expressions, which is generated by human beings and disseminated or amplified by software. Arguably, in an era of government surveillance and surveillance capitalism, anonymous expressions should be recognized as part of freedom of expression.

A third reason the court gave for the decision to reject the petition at this early stage of the proceedings was the petitioners’ failure to establish that the individuals affected by the takedowns were subject to the territorial jurisdiction of the state of Israel and protected by its Basic Laws. It seems that the court assumed that given the predominance of requests involving national security offenses, it stands to reason that many individuals who posted the relevant unlawful content resided outside Israel. Here, too, the court’s approach raises several difficulties. First, government measures aimed directly at removing online content posted by a specific individual located outside the territory of the state might entail responsibility, according to certain interpretations of international human rights law, for potential human rights violations; and the position of Israeli constitutional law on the matter is not entirely clear. Second, it may be presumed that some of the foreign nationals residing outside Israel live in the occupied territories, over which Israel exercises jurisdiction under mainstream interpretations of the International Covenant on Civil and Political Rights. Finally, if the offenses were committed outside the territory of the state of Israel by foreign nationals who are not protected by its Basic Laws, does the Cyber Unit even have the legal authority to request a takedown as a derivative of its authority to enforce the laws of the state of Israel?

Another fault that the court found with the petitioners was that the petitioners did not designate the online platforms themselves as responders to the petition. Chief Justice Esther Hayut, who wrote a short concurring opinion, argued that this was a fatal flaw of the petition, which, along with the lack of evidence of a specific infringement of a constitutional right, should have led to its rejection. Justice Hanan Melcer, who wrote the leading opinion, was less harsh on the petitioners, yet he was of the view that the absence of online platforms from the proceedings rendered it more difficult for the court to reject the Cyber Unit’s claim that the platforms exercise independent discretion and that, as a result, there is no direct relationship between the takedown requests and any constitutional right infringement. He maintained, in this regard, that the online platforms, and not the government, are ultimately responsible under this division of labor for any rights infringement that might have occurred as a result of takedowns.

The court’s position on this point is not clear to us. If, as the court accepted, the Cyber Unit possessed the ability to impose regulations on online platforms operating in Israel, it is unclear how additional information on the manner in which the platforms consider takedown requests would have changed the right-infringing potential of these requests. Israel does not have a Section 230-type arrangement that would provide online platforms immunity for civil liability for hosted content; hence, requests made from a unit operating from within the State Attorney’s Office might have been regarded by the platforms as an offer that they could not refuse without assuming a concrete legal risk for civil or even criminal liability. It is therefore questionable whether failing to designate the online platforms as responders in a public law case alleging infringements of constitutional rights by the Cyber Unit was indeed a serious procedural flaw. Moreover, the shifting of ultimate responsibility by the court from the government to the online platforms ignores the positive obligations of the state to ensure human rights (and constitutional rights) protections. Requesting the platforms to infringe rights cannot, by any stretch of the imagination, be viewed as compatible with this positive obligation.

Interestingly, the decision provides insight into the process of legitimizing self-regulation facilitated by the creation of the Facebook Oversight Board. The court noted that the ability to appeal takedowns before the board provides a remedy for rights infringements that might have occurred. This assessment of the impact of the board on the constitutional law analysis ignores, however, the fact that the board is capable of reviewing only a small number of borderline cases chosen out of many thousands of review requests. It cannot be regarded as an effective remedy to the millions of users whose content is removed from platforms on a daily basis.

Finally, the decision identified—in response to an amicus curiae brief submitted by the Freedom of Information Association—certain procedural shortcomings in the operations of the Cyber Unit, including lack of a clear and transparent process of communicating with the online platforms and a failure to keep a full record of the different takedown requests made. In light of these shortcomings, as well as the broader concerns discussed in the decision, the court recommended that the power to issue takedown requests be specifically set out in new primary legislation and that the practices of the Cyber Unit be subject to regular monitoring.

Conclusion

The Adalah v. Cyber Unit decision is the first comprehensive attempt by Israel’s High Court of Justice to review the constitutionality of government takedown requests pertaining to illegal content found on online platforms. The decision is premised, however, on a series of questionable factual and legal assumptions relating to the interpretation and application of constitutional rights in an online environment. Reliance on such assumptions suggests perhaps a strong interest by the court to rid itself of the “hot potato” of reviewing an administrative practice, which is useful but legally dubious. Ultimately, the decision created more problems than it solved: It accepted that the requests constitute a government act requiring statutory authorization, but it suggested for some reason that the platforms, not the government, are responsible for infringements following such requests. The court’s position on the scope of constitutional protections and the evidentiary basis required to trigger such protection is also unclear and appears to suffer from logical inconsistencies. Still, the decision leaves the door open for future petitions against the practices of the Cyber Unit, which may lead to the court striking down the power to make takedown requests without specific legislative authorization.