Ed: You open your recent Policy and Internet article by noting that “the modern treasury of public institutions is where the wealth of public information is stored and processed” … what are the challenges of government use of cloud services?

Kristina: The public sector is a very large user of information technology but data handling policies, vendor accreditation and procurement often predate the era of cloud computing. Governments first have to put in place new internal policies to ensure the security and integrity of their information assets residing in the cloud. Through this process governments are discovering that their traditional notions of control are challenged because cloud services are virtual, dynamic, and operate across borders.

One central concern of those governments that are leading in the public sector’s migration to cloud computing is how to retain unconditional sovereignty over their data — after all, public sector information embodies the past, the present, and the future of a country. The ability to govern presupposes command and control over government information to the extent necessary to deliver public services, protect citizens’ personal data and to ensure the integrity of the state, among other considerations. One could even assert that in today’s interconnected world national sovereignty is conditional upon adequate data sovereignty.

Ed: A basic question: if a country’s health records (in the cloud) temporarily reside on / are processed on commercial servers in a different country: who is liable for the integrity and protection of that data, and under who’s legal scheme? ie can a country actually technically lose sovereignty over its data?

Kristina: There is always one line of responsibility flowing from the contract with the cloud service provider. However, when these health records cross borders they are effectively governed under a third country’s jurisdiction where disclosure authorities vis-à-vis the cloud service provider can likely be invoked. In some situations the geographical whereabouts of the public health records is not even that important because certain countries’ legislation has extra-territorial reach and it suffices that the cloud service provider is under an obligation to turn over data in its custody. In both situations countries’ exclusive sovereignty over public sector information would be contested. And service providers may find themselves in a Catch22 when they have to decide their legitimate course of action.

Ed: Is there a sense of how many government services are currently hosted “in the cloud”; and have there been any known problems so far about access and jurisdiction?

Kristina: The US has published some targets but otherwise we have no sense of the magnitude of government cloud computing. It is certainly an ever growing phenomenon in leading countries, for example both the US Federal Cloud Computing Strategy and the United Kingdom’s G-Cloud Framework leverage public sector cloud migration with a cloud-first strategy and they operate government application stores where public authorities can self-provision themselves with cloud-based IT services. Until now, the issues of access and jurisdiction have primarily been discussed in terms of risk (as I showed in my article) with governments adopting strategies to keep their public records within national territory, even if they are residing on a cloud service.

Ed: Is there anything about the cloud that is actually functionally novel; ie that calls for new regulation at national or international level, beyond existing data legislation?

Kristina: Cloud services are not meant to recognize national frontiers, but to thrive on economies of scale and scope globally. The legal risks arising from its transnationality won’t be solved by more legislation at the national level; even if this is a pragmatic solution, the resurrection of territoriality in cloud service contracts with the government conflicts with scalability. My article explores various avenues at the international level, for example extending diplomatic immunity, international agreements for cross-border data transfers, and reliance on mutual legal assistance treaties but in my opinion they do not satisfyingly restore a country’s quest for data sovereignty in the cloud context. In the EU a regional approach could be feasible and I am very much drawn by the idea of a European cloud environment where common information assurance principles prevail — also curtailing individual member states’ disclosure authorities.

Ed: As the economies of scale of cloud services kick in, do you think we will see increasing commercialisation of public record storing and processing (with a possible further erosion of national sovereignty)?

Kristina: Where governments have the capability they adopt a differentiated, risk-based approach corresponding to the information’s security classification: data in the public domain or that have low security markings are suitable for cloud services without further restrictions. Data that has medium security markings may still be processed on cloud services but are often confined to the national territory. Beyond this threshold, i.e. for sensitive and classified information, cloud services are not an option, judging from analysis of the emerging practice in the U.S., the UK, Canada and Australia. What we will increasingly see is IT-outsourcing that is labelled “cloud” despite not meeting the specifications of a true cloud service. Some governments are more inclined to introduce dedicated private “clouds” that are not fully scalable, in other words central data centres. For a vast number of countries, including developing ones, the options are further limited because there is no local cloud infrastructure and/or the public sector cannot afford to contract a dedicated government cloud. In this situation I could imagine an increasing reliance on transnational cloud services, with all the attendant pros and cons.

Ed: How do these sovereignty / jurisdiction / data protection questions relate to the revelations around the NSA’s PRISM surveillance programme?

Kristina: It only confirms that disclosure authorities are extensively used for intelligence gathering and that legal risks have to be taken as seriously as technical vulnerabilities. As a consequence of the Snowden revelations it is quite likely that the sensitivity of governments (as well as private sector organizations) to the impact of foreign jurisdictions will become even more pronounced. For example, there are reports estimating that the lack of trust in US-based cloud services is bound to affect the industry’s growth.

Ed: Could this usher in a whole new industry of ‘guaranteed’ national clouds..? ie how is the industry responding to these worries?

Kristina: This is already happening; in particular, European and Asian players are being very vocal in terms of marketing their regional or national cloud offerings as compatible with specific jurisdiction or national data protection frameworks.

Ed: And finally, who do you think is driving the debate about sovereignty and cloud services: government or industry?

Kristina: In the Western world it is government with its special security needs and buying power to which industry is responsive. As a nascent technology cloud services nonetheless thrive on business with governments because it opens new markets where previously in-house IT services dominated in the public sector.

Read the full paper: Kristina Irion (2013) Government Cloud Computing and National Data Sovereignty. Policy and Internet 4 (3/4) 40–71.

Kristina Irion was talking to blog editor David Sutcliffe.

Ed: European legislation introduced in 2011 requires Member States to ensure the prompt removal of child pornography websites hosted in their territory and to endeavour to obtain the removal of such websites hosted outside; leaving open the option to block access by users within their own territory. What is problematic about this blocking?

Authors: From a technical point of view, all possible blocking methods that could be used by Member States are ineffective as they can all be circumvented very easily. The use of widely available technologies (like encryption or proxy servers) or tiny changes in computer configurations (for instance the choice of DNS-server), that may also be used for better performance or the enhancement of security or privacy, enable circumvention of blocking methods. Another problem arises from the fact that this legislation only targets website content while offenders often use other technologies such as peer-to-peer systems, newsgroups or email.

Ed: Many of these blocking activities stem from European efforts to combat child pornography, but you suggest that child protection may be used as a way to add other types of content to lists of blocked sites – notably those that purportedly violate copyright. Can you explain how this “mission creep” is occurring, and what the risks are?

Authors: Combating child pornography and child abuse is a universal and legitimate concern. With regard to this subject there is a worldwide consensus that action must be undertaken in order to punish abusers and protect children. Blocking measures are usually advocated on the basis of the argument that access to these images must be prevented, hence avoiding that users stumble upon child pornography inadvertently. Whereas this seems reasonable with regard to this particular type of content, in some countries governments increasingly use blocking mechanisms for other ‘illegal’ content, such as gambling or copyright-infringing content, often in a very non-transparent way, without clear or established procedures.

It is, in our view, especially important at a time when governments do not hesitate to carry out secret online surveillance of citizens without any transparency or accountability, that any interference with online content must be clearly prescribed by law, have a legitimate aim and, most importantly, be proportional and not go beyond what is necessary to achieve that aim. In addition, the role of private actors, such as ISPs, search engine companies or social networks, must be very carefully considered. It must be clear that decisions about which content or behaviours are illegal and/or harmful must be taken or at least be surveyed by the judicial power in a democratic society.

Ed: You suggest that removal of websites at their source (mostly in the US and Canada) is a more effective means of stopping the distribution of child pornography — but that European law enforcement has often been insufficiently committed to such action. Why is this? And how easy are cross-jurisdictional efforts to tackle this sort of content?

Authors: The blocking of websites is, although questionably ineffective as a method of making the content inaccessible, a quick way to be seen to take action against the appearance of unwanted material on the Internet. The removal of content on the other hand requires not only the identification of those responsible for hosting the content but more importantly the actual perpetrators. This is of course a more intrusive and lengthy process, for which law enforcement agencies currently lack resources.

Moreover, these agencies may indeed run into obstacles related to territorial jurisdiction and difficult international cooperation. However, prioritising and investing in actual removal of content, even though not feasible in certain circumstances, will ensure that child sexual abuse images do not further circulate, and, hence, that the risk of repetitive re-victimization of abused children is reduced.

Read the full paper: Karel Demeyer, Eva Lievens and Jos Dumortier (2012) Blocking and Removing Illegal Child Sexual Content: Analysis from a Technical and Legal Perspective. Policy and Internet 4 (3-4).

Karel Demeyer, Eva Lievens and Jos Dumortier were talking to blog editor Heather Ford.