FINISHED COPY SEVENTH ANNUAL INTERNET GOVERNANCE FORUM BAKU, AZERBAIJAN SUSTAINABLE HUMAN, ECONOMIC, AND SOCIAL DEVELOPMENT 7 NOVEMBER 2012 11:00 AM SESSION 172 CLOUDY JURISDICTION: ADDRESSING THE THIRST FOR CLOUD DATA IN DOMESTIC PROCESSES * * * This text is being provided in a rough draft format. Communication Access Realtime Translation (CART) is provided in order to facilitate communication accessibility and may not be a totally verbatim record of the proceedings. * * * >> RAPPORTEUR: Thank you, everyone, for coming to the session today. I want do make sure everyone has headphones. We are trying to resolve some trans-border threats to civil liberties posed by the move to the cloud. If a baseline of privacy protection can be assured, concerns over limiting data flows on the basis of jurisdiction will be alleviated. This panel will be divided into two parts. The first part will discuss some of the challenges raised by the cloud environment for traditional civil liberties paradigms. The discussion in part two will be solution-driven. What rules can be put in place at the international or natural level to alleviate the heightened risk to privacy and other civil liberties raised by the cloudcentric model. To my right I will invite first Mr. Bertrand LaChapelle who is the Programme Director at International Diplomatic Academy. He will raise the concerns or the challenges of jurisdiction on the cloud. Welcome, Bertrand, thank you so much for coming. >> BERTRAND LaCHAPELLE: Thank you. So my name is Bertrand LaChapelle. I am at the International Academy leading a programme called Internet and Jurisdictions and if you have looked in your bag you may have seen this brochure and the invitation to the workshops we hold on Thursday and Friday on those topics. I'm glad the opportunity is given to highlight a few points regarding jurisdiction. In the last year and a half, this word "jurisdiction" has popped up in discussions where it was relatively rare before. The challenge that we are witnessing with the tension between the horizontal cross-border nature of the Internet and the vertical system of national jurisdiction is growing. I want to make a clarification, first of all. The Internet is technically borderless. It is not based on a geographic architecture. However, this does not mean there are no borders and jurisdictional borders on the Internet. When you move from a website in .com to a website in .cn you are crossing borders even if you don't understand it. You may be crossing from France to Belgium without any control, nonetheless, you are under different jurisdictions. The key problem is that therefore we should talk about the Internet and the Internet spaces, the geography of the jurisdictions, do not map one-on-one the physical jurisdiction of the different countries. Without getting into details, there are extensions of the power of sovereignty outside, and vice versa; there are situations where the citizens of one country are actually under the quasi-jurisdiction of another because of the types of Internet applications they use, because of the types of the domain names they have bought. To give you one concrete example, you are probably familiar with, when the website was seized by the Homeland Security arm, the ICE arm of Homeland Security because it was bought by a Spanish operator through an American-based registrar, there was a de facto extension, currently legal in the current architecture of the natural sovereignty over the territory of another country. And there are many other examples of spillover effects like, for instance, when the filtering that is applied to some ISPs in India filter down to Oman because of the arraignments between the operator in Oman and the telco operator in India. I don't want to get into detail but I wanted to highlight we are witnessing a situation where the exercise of national sovereignty has potential trans-boundary impacts on another territory. The problem is that for a lot of platforms and services that use the Internet and target users around the world, there is a tension between their terms of service and the national laws of all the countries they operate in. Without getting into details, what I want to highlight is the danger or the situation that we are witnessing today is not the type of situation we had in the past whereby sovereignty was working by separation. It is still about the clear frontiers where the sovereignty of one country is on one side and another country is on the other side. What is at stake is the management of sovereignty and jurisdiction over shared spaces. And overlayering of jurisdictional competence rules that take into account -- here again, not getting into too much detail -criteria that are the location of the user, the country of incorporation of the platform of operator, the location of its servers, and even the type of domain name that this operator is using or where it bought the domain name in question. So to bring this larger picture of overlapping jurisdictional problems to the question of cloud computing or cloud services, if you take into account location of servers to apply jurisdictional criteria, you have a new type of tension. The benefit is precisely the data is distributed on several servers that ideally are distributed around the world to provide the better quality of service. If you need to have an application of all different jurisdictional criteria based on where the service is, you destroy the very purpose of cloud. As an illustration of this problem, you may not know that Jimmy Wales, founder of Wikipedia, says that he does not put any server for Wikipedia in England because of the libel laws in Great Britain that would not be compatible with the way Wikipedia works. Wikipedia as a matter of fact is a typical example of the kind of problems that cloud computing may more than other actors experience with the patchwork of jurisdictional criteria. >> RAPPORTEUR: Thank you, Bertrand. Now we are going to have a discussion about recurring problems that have arisen in many comparable online contexts when it is related to the cloud. We have legal obligations. Some governments, some governments are discussing some new legal obligations to build in intercept capacity into Internet services, whether it's Canada, UK, United States, United States in previous years, and some other challenges regarding voluntary sharing of data when, for instance, even though they, the police will need a warrant to access communications, sometimes there are no -- there are immunity -- no liability in the statute is the sharing the data voluntarily. And if they need to share it, we should know the limits. There are a lot of challenges right now taking into account that lately we are using the cloud in our daily lives, we are relying more and more on cloud services, and all the data that in the past was in a house or an office are now moving to the computers. It's not any longer in the hand of each of us in our house but in the hand of third-party providers. So we are going to invite Ian Brown. Ian Brown, where are you? He is Senior Research Fellow at the Oxford Internet institute. He is a well-known academic; he has also written recently a report for the global network initiative on cloud computing. Thank you for coming, Ian. >> IAN BROWN: Thanks, Katitza. I'll say a bit about the UK in this first part because I think the UK is really a leading government on Internet surveillance in ways perhaps the civil liberties community would not like. Let me briefly, Bertrand, on the libel side, that libel question that Jimmy Wales raised is very interesting because in a way the UK libel law is an illustration of almost preInternet, very sweeping claims of jurisdiction within the UK that in an Internet era are very problematic. And I don't like to disappoint Jimmy Wales but the fact the servers are not in the UK would not prevent the UK courts from finding a connection to English law! (Laughter) Although we are trying to reform our libel laws right now. Lawful intercept in particular, let me say a few words. Current government actually current government and the previous government have been developing for several years now at new lawful interception requirements on communication service providers. Like many other countries, including the U.S., we have had for a number of years requirements that telecommunications service providers have a lawful intercept capability. That is to say, if a government agency comes to an ISP or phone company with a lawful warrant in the UK case, not a court order, a warrant signed by the Secretary of State, cabinet member, authorizing an interception, then the Internet service provider has to be able to undertake that interception. Alongside that, again, like many other jurisdictions, we have laws requiring Internet service providers to keep basic communications records about what their users are doing online. When they were connected to the Internet, who they have been communicating with via e-mail and the case of phone companies via mobile telephones and also some specific information about location of the phones that must be recorded. That is an EUwide requirement under the data retention directive although that was driven by the UK and the UK had broader retention laws before the European data retention directive came into place. The UK law enforcement and intelligence agencies are now complaining that the data that is retained about people's Internet use is not broad enough. As people move from traditional e-mail to things like social networks and to communicate instant messaging, even online virtual environments and online games, those agencies are concerned that they will lose the capability to put under surveillance people that are using those mechanisms to plan and commit criminal acts. Therefore, the government, current government, has proposed a very sweeping bill in Parliament which is currently being looked at by a special select committee of Parliament that would really potentially broaden out those requirements for intercept and this particularly storage of this communications data about what people are doing. It could even go as far as under the text of the bill giving the government the ability to require not just Internet service providers but actually pretty much anyone providing any kind of communication service online. To some extent it could be argued even down to an individual that was running something like a tour node on a home PC to record this communications data and also to facilitate the recording of communications data and intercept. Right down to the level of specific algorithm, cryptographic algorithms that a system could use. Mobile phones of course encrypt voice, tour is encrypting data packets so very broad-ranging bill. Lot of criticism from the civil liberties community and it's not clear whether this bill will pass into law as is or will be amendments in Parliament or whether the good afternoon as has happened in the past with this type of legislation may have to step back, withdraw the bill and introduce something more limited in the future. I just will say a few other things about the UK regime that I think are interesting for this discussion. A big issue as Katitza mentioned about privacy of people's communications records especially but more broadly data about activities. In many countries, including the UK, government agencies can ask companies that hold that data to voluntarily provide it to the government agency. In the UK the Data Protection Act specifically allows for purposes related to criminal investigations, a broader range of purposes than that for data control is to voluntarily provide data to the government and in trying to work out as I did for an academic article earlier this year which you can find online called Government Access to Private Sector Data in the United Kingdom. I suspect, although I can't confirm because there are not procedures by which these voluntary provisions of data have to be notified, say, to the information commissioner, the data protection regulator for the UK, certainly not made public or notified to the people whose data is being handed over, there is no central point at which you can find information about that. I suspect there is a lot of this going on, a lot of data flows from the private sector to UK government agencies that are not transparent and notified to individual users. I think that's one big problem. A second problem is that the UK surveillance regime in general is very untransparent. People are not notified if they have been the subject of investigation after the investigation has closed, which would be a good way of trying to prevent abuse. The normal courts do not by and large have jurisdiction over how these powers are used. Intercept evidence is not allowed in court cases because the intelligence agencies worry that would reveal their methods and sources. So in day-to-day hearings courts are not looking at the evidence and whether it was lawfully obtained. There is a special tribunal that exists to investigate alleged abuses of interception capabilities by the intelligence agencies. However, they hear very few cases because how would people ever know if data about them or their communications have been intercepted if they aren't notified or there is not some other organized way of that notification happening? In practice, that tribunal has only I think in -- it has heard a few hundred cases during its lifetime of about I think a decade and it has only found in favor of the complainant in I think six cases the last time I looked at that. There is limited information about their decisions, of course; they won't publish full details about what was alleged. Then finally I think again interesting in the UK for the broader discussion, the ability for one-stop shop access by government agencies to data about people's online activities. We certainly have the case in the UK that accredited police agencies are able on a particularized but automated basis to access communications data about people. So records of what people have been doing online today, that's reasonably limited, that communications data, just basic data about their -- when they were online, their subscriber data, names and addresses, ISP holds, who they were talking to by e-mail. A big problem if you massively broaden out the scope of this communications data as the communication data bill would, then you are talking perhaps about a lot of other information about people's online activities and social networks and online games, more geolocation information in total which will paint a very detailed picture of an individual's life. I think there's now a false dichotomy in many countries' laws between contents of the communication, what's in the email, which in many countries is strictly protected, versus this communications data, these records about what people are doing which are becoming richer and richer and broader and are accessed under much less strict safeguards. That is something that the cloud will make even worse and I think we have to think very carefully about and we'll come back to solutions later in the session. >> RAPPORTEUR: Thank you, Ian. It's interesting the comments you make about the voluntary sharing. I believe that not only that's a bigger problem for especially small companies who might not have the capacity or resources to be able to challenge those requests or that political pressure from the government. Okay. So now we would like to learn a little more about the challenges that are being -- happening in India. We have invited for that Eleni Hickock, policy associate for the Center for Internet and Society in India. >> ELENI HICKOCK: Hello, I'm Eleni Hickock and I work for the Center for Internet and Society based in Bangalore. It is an NGO so I'm representing the Civil Society perspective of these issues. I think what we have seen from the Indian government and their response to the cloud and some issues that have arisen is that traditional forms of lawful access are not applying to the cloud because of the distributed issue and the jurisdiction issue. I think the Indian government looked toward (indiscernible) to solve that but that is not working and many people in India are unhappy with that process, so I'm sure you guys have heard about the RIM scandals so the Indian government has been asking for blanket access and encryption keys from different server providers to monitor these communications and monitor the data that is happening and going to other countries. So far the government has asked for RIM keys and finally RIM set up in Mumbai and now again RIM has suggested that intelligence agencies in India have the capability or create the capability to intercept those communications as well. I think there are a lot of factors that play into this response from the Indian government and a lot of factors in India that also complicate this issue. So for one India does not have a comprehensive privacy legislation. That creates vulnerabilities for data stored in the cloud and stored in India in general. I think also service providers in India are decentivized from not providing and not complying with lawful access requests or access requests from the government because of heavy penalties the government puts on non-cooperation. So, for example, service providers can be put in jail or imprisoned for seven years for non-cooperation. This has actually changed over the years. The Telegraph Act first said they would be penalized with six months in prison and now the ITA says seven years in prison. At the same time intelligence agencies are not held liable for illegal interception that happens. There are issues of liability and also there's no incentive for service providers to really protect information and not hand it over to intelligence agencies. I think you have also issues, cultural factors as well, feeding into the government fear and why they want to access all this information. This might be very much a developing country issue that needs to be taken into consideration when we look at principles. There are internal threats that are very real and India has a huge population. Lots of diversity in religions and ethnicities and these are all factors the government tries to balance and it gets augmented when suddenly the information is stored on the cloud. I think you have seen some emerging trends from the Indian government so there is emerging legislation coming out that is asking for a broader retention of data and longer retention of data and at the same time, there is unclear authorization as to who can access that data. You have authorization standards lowering, broader retention standards happening, and this is complicating civil liberties, as Katitza pointed out. I think there is also a problem with implementation in India. Though there are safeguards that do exist in the Indian regime these are not always followed and leaks happen and data is not handled as it's supposed to be handled and you have problems with chain of custody. You saw the (indiscernible) tapes; that's one example of these issues. I think that's a short summary of some of the issues that developing countries have around this issue of lawful access and some of these issues are augmented or exacerbated by the cloud, I think. >> RAPPORTEUR: Thank you, Eleni. We have the Civil Society to be able to challenge surveillance because the programs are sure that -- shrouded in secrecy because individuals are never made aware because of a standing issues, et cetera. Some companies like Twitter and Google have been trying to shed light on these issues and we have invited today Marc Crandall from the global compliance team in Google to shed light a little about the problems with the transparency report and whatever reactions from the panel. >> MARC CRANDALL: I think at Google one of our primary concerns, particularly within the United States, is reforming our existing government access laws when it comes to online information. For example, in the United States we have something called the Electronic Communications Privacy Act, passed in the 1980s, which involves the steps that governments have to take in order to obtain information about users online. We think that those laws, for example, need to be updated so that the protections that exist for information that is in your home also exist for the same type of information that would otherwise be stored in the cloud. In that respect, we take a leading role in what's called the Digital Due Process Coalition, which is a reform advocacy coalition involving this type of law which we just talked about. Essentially we'd like to see these laws modernized in a number of ways, better protection of data stored online, similar to what I just mentioned. We feel the government must first get a search warrant before obtaining private communications or documents stored online. Government requires process in order to get this information typically but we would like to see the same for online storage as what would otherwise be required for the government to go into your house. We'd like to see better protection regarding location. We want to see better protection regarding location privacy. We think the government should also get a search warrant before contracting location of your cell phone or other mobile communications device. We think it should be updated in that regard as well. We all use -- a lot of us, I should say, use devices that have location capabilities and that should require equal protection. We'd like better protection against monitoring of when and with whom you would communicate. The government must demonstrate to a court we feel that the data it seeks is relevant to a criminal investigation before monitoring when and with whom you communicate using e-mail, instant messaging, text, the telephone, anything. Finally, we think we need better protection against bulk data requests. We talked about the growing corpus of information online. We feel the government needs to demonstrate to a court that the information it seeks is needed for a criminal investigation before it can obtain data about an entire class of users. This is what we'd like to see with regard to the United States and reform of our own regulations. >> RAPPORTEUR: Thank you. Thank you, Marc. I would like to see if there are any questions from the remote moderator, online participation from the remote moderator, please. No? (Silence) Are there any questions online? >> MATT ZIMMERMAN: No, no questions online yet. >> RAPPORTEUR: Okay. I would like to see if there's one from the law enforcement community or someone who would like to make a comment on the panel so far a question. Please, the lady on the end. (Pause) The microphone, please, for the lady on the end. >> Thank you. We are hearing all the time from the Civil Society that some ideas like go and rectify the laws -- it's fine? >> RAPPORTEUR: It's fine. >> We know when legislators want to make laws they have to understand the subject of the law, the situation, to be regulated. So what are technical people, what are technology people and Civil Society saying to legislators? Okay? We are all the time nagging and saying, oh, civil liberties are at risk, we are all the time controlled, we are tracked, why don't we tell legislators how to do, to get rules convenient to the new environment which is the Internet to the needs to protect national security and then to other needs to feel free and to feel that way all the time? I mean, we have to give them factual things. >> RAPPORTEUR: I would like to give the floor to Wendy Seltzer from the technical community to give a brief intervention. >> WENDY SELTZER: One of the things that technologists are often called upon to help with is to explain the possibilities and impossibilities and so sometimes there are difficult questions asked to which it's not possible to give a clean technological answer that says yes, if only you take these steps you will have the solution to all of your law enforcement problems or, on the other side, all of your civil liberties problems. These are messy issues. What we can do is to help to elucidate the challenges and some of the solutions that are better and worse for solving those problems. So we can note, for example, that because of the difference in jurisdiction and fluidity of data in the cloud, if people don't have protection in one place for the activities they want to engage in, they may simply go someplace else where their data will not be subject to the same retention and tracking rules and circumvent the kinds of protections that someone meant to put in place by imposing that surveillance. So that a solid protective regime for people trying to store data online can be helpful to everyone who is trying to achieve a better relationship there. It's good for companies because they can sell a stronger product to their consumers. It's better for law enforcement because at least it keeps the activity jurisdiction rather than sending it elsewhere where they have even less control and it's better for those seeking privacy because it assures due process and transparency of the rules relating to information collection and its use. >> RAPPORTEUR: Thank you, Wendy. Now we call on Bruce Schneier who is Chief Security Technology Officer of VT from the private sector, also from the United States. >> BRUCE SCHNEIER: Hi, good morning. I want to make three quick points. One, it has been said many times that data is moving to the cloud. In general, the problem I think we're dealing with is that people are losing control over their computing. It's happening from two different dimensions. On the one hand, our data is moving to the cloud where the regulations are not the same as if the data was in our house. Perhaps held by a third party, perhaps held in a different country, perhaps across -- it crosses borders and there are a lot, these jurisdictional issues are very difficult and things we're wrestling with. The other end of this is that we're losing control of our end devices as well. I mean, I'm using an iPhone and I have much less control over what I can put on this device than I do my computer. I can't put an anti-virus, I can't even write a file erasure programme. Updates are largely opaque to me. And this is happening a lot, whether it's smartphones or tablets or e-book readers or gaming consoles or cameras. These devices are increasingly Internet-ready and increasingly opaque to the user. In both cases, there is a lot of issues of control and really of trust. That we are trusting whoever makes either the end user devices or the in-cloud data stores to protect us by obeying the law, keeping the data within borders and there are technical solutions to these things. But it's not clear that the companies that build these devices really want them. There is a lot of value to keeping data unencrypted in a cloud provider, cloud provider wants to mine that data for advertising purposes and or for beneficial purposes for you. I think that is the big trend here that we're trying to fight. We're now moving to a level of sophistication in computing where things are moving out of our control but that has legal and jurisdictional implications. That's the first big trend. The second is that as we're learning, governments are discovering the Internet. We're using it for more and more of our socialization, of our business, commerce, more and more of our lives are moving onto the Internet so we're seeing a more government scrutiny. The question is why can't the technologists tell the lawmakers what to do? The problem is the lawmakers don't want to hear that. The lawmakers are seeing balances that they set 10 or 20 or 30 years ago being upset by this new technology. What they want is to move the old regime, whatever it was, into this new technology, which often is impossible, because it doesn't work the same. But that does not stop lawmakers from trying. Lawmakers are under pressure from two areas; from police forces, who want access to data for various reasons; and they're under pressure from industries. They want things to remain the same. In the United States and elsewhere we're seeing enormous fights by industries who make businesses on copyrights trying to force the Internet to be just like records and tapes and other physical objects. It's not working. So what we're seeing, I think, I mean, I see it in the U.S. and certainly elsewhere, is very heavy-handed Internet regulation that does not really take into account the subtleties of the Internet and I think that causes more problems than it solves. Last year in the United States we had a debate over an Internet kill switch. That debate takes many forms. I think of it as a big red button on Obama's desk. But however you think about it as an security engineer that's an utter disaster, but as a police force who wants to say, shut down the phone system, it's the same thing they had before. Convincing them it's difficult is hard. Those are what I think are the two megatrends that really affect this issue, and I'll stop there. >> RAPPORTEUR: Thank you, Bruce. Now we have Sophie Kwasny from the Council of Europe. Thank you, Sophie, for coming to our meeting today. >> SOPHIE KWASNY: Thank you. Sophie Kwasny. Yes, I work for the Council of Europe in charge of data protection but I work for a wider division which is the cybercrime and data protection division and I think that the fact that those two issues have been brought together recently in the Council of Europe is a good signal. We have heard that many of the issues discussed today infringe upon civil liberties, right to privacy in particular, so bringing them together at the Secretariat level, it's enabling us to really work closer together on those issue and so I'm very happy even with the data protection background to come and mention some of the issues raised by the cloud. So it was said there is a regulatory framework that's been there for years, been working, the problem with the cloud, cloudy jurisdiction, can be foggy, murky, the fact the boundaries we knew are more and more blurred. They are blurred on a number of levels. First is that the law enforcement side access by law enforcement is shifting in some places to surveillance intelligence and traditionally if some of the safeguards we have been putting in place in the Council of Europe have begin applicable to the law enforcement, it's true for intelligence side it's always a bit more difficult. So making a clear distinction between both is important and in practice it is more and more difficult. Another type of blurring is about the data. Those regulatory frameworks, they define the types of data that can be accessed. I will use the terminology of the Council of Europe which is the cybercrime convention terminology. We see that it can be accessed to traffic data but when you are accessing traffic data, are you solely accessing that or also content data? There again the frontier between those types of data is less clear than it used to be in the past. Is data at rest? Is it stored on a computer and can be accessed under a search and seizure or it data entrance mission which then should be covered by other types of mechanisms? We heard about the voluntary transmission of data. Indeed under the convention of cybercrime detecting access to publicly available data. This can be accessible with no further authorization. And when there is access control restrictions, lawful and voluntary consent must be obtained from someone with lawful authority to disclose data. It's this voluntarily mechanism, the question being who is consent to go that? Who has this legal authority? Is it the cloud user? Is it the cloud provider? Finally, blurring of the form of cooperation. It's from formal to informal, from legality of framework to requests covered by infinite terms of service of the cloud providers. The last point is the blurring of frontiers that was mentioned before, jurisdiction questions. Which law applies? Which safeguard applies? And the notion of consent is also understood differently around the planet so how do we apply that? Those are basically for me the issues at the moment. And if you allow, we'll tackle later the solutions that can be proposed. Thank you. >> RAPPORTEUR: Thank you, Sophie. I would like to take four or five questions from the floor if there are any. One. >> Good afternoon. I'm (inaudible) and I represent (inaudible). Good afternoon. I represent Internet Service Providers Association of India. I happen to sit on the executive council. I have a question for the panelists just to give a brief background. An Indian regulator has called for an approach paper on concerns about data and cloud computing. I'd like to hear from the panelists today on are there any best practices or guiding principles we as service providers should be giving input to the regulator on how best we should approach these concerns about cross-border data protection issue and jurisdiction issues? Is that something we could, as an approach to the government, suggest? Because the government is inviting service providers to come to the table and discuss all these issues. Thank you. >> RAPPORTEUR: Another question. Just to clarify a little. We are going to the second part of the panel about solutions-driven and your question will feed perfectly into that. I will wait for the panelists to get to the second part before they answer to you. We wrote your question. Is there any other questions? >> In order to be a little contentious, one of the things here we were talking about is changing of defaults. In the sense that if everyone is using SSL and PGP, then that significantly changes the default from the age of the telegraph, interception then was easier. So when only those people who wanted to encrypt would use code language in there, so how do we address this issue? Now, this is something that law enforcement agencies are actually quite concerned about. It's not about whether person can because they can always do that, they have always done that, but the issue of changing of defaults is a problem that, and so I would like to kind of reiterate what the lady from the back said. We are bringing up all kinds of issues with privacy but I think we have to provide good ideas about security as well, not for individual security but how law enforcement agencies should and could go about it which does not rely solely on the idea that judges can clear it because judges can be people who are not well-versed with the constitutions of the country, people who, you know, executive magistrates who were policemen in India can be counted as judges for these purposes as well. That can't be the single-point solution. >> Just specifically on that, well on on the two points, I think, first of all, the rule of law is the best that we have and if you have problems in individual jurisdictions with the judiciary that's something that needs to be addressed within the judiciary and level which authorizations can be given. The encryption question is really interesting because this came up a lot in the UK bill because of course what the government would like to happen is Internet service providers intercept using deep-packet inspection equipment all traffic flowing to servers where the government does not affect -effectively trust that that international provider will under the right circumstances provide some of this communications data, so the ISP themselves within the UK jurisdiction can provide and what happens if the traffic flow is encrypted? This comes back to the question from the lady from the back. This question caused enormous confusion to the government. They -- you would not expect MPs to read Bruce's wonderful book on 500 pages of cryptographic algorithms but the concept, it won't mean anything, seemed to be very hard to get over to the government in the UK. But they finally got it and they came up with some extraordinary solutions like we'll record everything for six months and then if we're interested in the traffic we'll go back to the provider and get those cryptographic keys and decrypt it, which was remarkably impractical. We'll come back, as Katitza said, to better solutions but the debate has been as confused in the UK as India, I think. >> RAPPORTEUR: Thank you. Any other questions? Please, the lady first, white lady, and then the lady in the back. >> I have a question for Bruce Schneier. I want now to know how do you differ cloudcentric security from traditional and what character makes -- (indiscernible) -- in cloudcentric? What characters make security issue cloudcentric? >> BRUCE SCHNEIER: The main difference between security on your desktop and security in the cloud is you don't often have access to the security controls. Especially as you move to cloud computing where you expect the cloud provider to do actual work for you. So if you are using Dropbox where you are just storing files, very simple cloud service, that's relatively easy for me to secure, encrypt files, move them. I don't care where Dropbox puts them, in what country, how they move them, they're not doing anything with that. Contrast that on the other end with something like Facebook. Facebook is only useful if that company has access to the actual data I post. They are in charge of who sees it, how it's used, how it's displayed. I have no ability to secure that data. I don't even know what operating system Facebook uses, let alone being able to audit their systems or mandate certain controls. So as we saw two weeks ago when they made a mistake and user data was visible to other users they may not want, we don't even understand what that mistake was. As the data moves out of your control, you have to trust the provider more. Even on your computer you to have trust your provider. I have to trust the vendors, but I still have some amount of control. I have my anti-virus, networking environments. As data moves to the cloud, as my computation moves to the cloud, I have much less control. E-mail, I have on Google servers, I have less control over the e-mail I have on my own servers. I'm trusting Google will probably secure my mail, only respond to lawful orders, I have to trust that company. That's the main difference. I have less ability to have control and less visibility to what controls are in place and I have to trust more. >> RAPPORTEUR: Mark and then Bertrand. >> MARC CRANDALL: That raises a very interesting point regarding lack of control in regards of trust. We are at an interesting crossroads I should stay in the development of cloud computing and the Internet people I think fully do trust some forms of online interaction, online banking, they trust that their bank account, their life savings, will be handled appropriately, for example, by the bank. And, for example, I have no problem believing that my life savings is represented by a number shown to me on the screen by my banking institution. I just happen to trust them. Why? Well maybe it's because we feel that they're a regulated entity. They have to be accountable to someone. Maybe it's psychological, something we've grown so used to over the years we just accept it. I remember the first time I deposited money using an ATM, deposited money and I really had a problem with that at the time. Now I'm used to it. So why do people trust putting or trust their life savings to online interactions with the bank but don't necessarily trust the disposition of their data to a cloud provider? What is the difference? What does the cloud provider have to do to earn that trust? That comes from anywhere from responding to third-party requests for your data that is being stored in the cloud, to security mechanisms to help prevent unauthorized access that is not due to rule of law. Breachers, hackers, and the like. So we have an interesting gray area from the Google perspective because we do provide enterprise, we provide to businesses, many of whom have their own regulatory compliance obligations in their own industries. Sometimes they're in certain parts of the world that have very strict privacy requirements like Europe with the U-data protection directive and they certainly have questions regarding law enforcement access, third-party access. What we do in that regard is try to provide as much control to the enterprise customers that are using cloud services as possible so we put it in their hands. We can't do it entirely of course but as much as we can. For example, in our situation for third-party requests for enterprise data, we want those requests to be handled by the customers themselves, not by Google. We don't want to be the compliance team for our customers. They will be in the best position to evaluate the process, determine what should be disclosed and what their options should be. When we can, we want to defer. In situations where it's not possible to notify affected customers, then, like all law-abiding customers that have to respond, we have an entire team of personnel dedicated to reviewing requests as appropriate to make sure they comply with not only the letter of the law but also the spirit of the law. If they don't, we have to fight it. So that's sort of the area we moved from with regards to consumer services like social networking to enterprise cloud services. The other thing I should mention is with regards to things like security. Providers often need to provide some sort of verification to enterprise cloud customers so it's one thing to say that we have great security. It's another thing to provide verification of that. Right? Because why should you believe us? Why should you just trust us? Because we're sitting here on a panel? So in the enterprise space what we often do is hire third-party auditors to come in to evaluate statements to make sure they're true. We also do things like attain a security standard, for example, ISO 27001 to show at the very least auditors have come in to verify what we're doing actually attains some sort of security standard. There are ways to bridge the gaps between complete lack of visibility and lack of control to empowering the user to have control and feel confident that the data they store online is as secure as data they store with the bank. >> RAPPORTEUR: Thank you. We are going to go to part two of the session. We will focus on how some of these problems can be addressed at the international level or national level by adoption of a set of principles and protections designed to meet the realities of online and specifically cloud services. The focus is on problem solution with the objective of providing concrete proposals for international- or national-level solutions. We will invite again Bertrand LaChapelle to just reply to one of the questions and make a statement. >> BERTRAND LaCHAPELLE: Thank you, Katitza. As a matter of fact, in a previous professional life between 2006 and 2010, I was the French representative for Internet Governance issues in the French Foreign Affairs Ministry. As probably the only person on this panel -- unless I don't know the bios enough -- that has had connections with governments I want to make one point nonetheless. And that is that there is a flip side for all discussions we have here which is that all of you are citizens and are concerned that the appropriate measures be put in place to identify the relevant information regarding cases where you really need to act, and I don't want to make the list. We all know this is the case. The big challenge is the challenge of balance, and the challenge we have established painfully between civil protection requirements and efficiency of law enforcement upon which we rely for some elements of security. This difficult balance that was achieved through sometimes fights in centuries in the traditional space is suddenly moving under our feet in the cyberspace. One of the reasons why is because if you think about the amount of data that is easily connected and easily analyzed, the thing that is at stake is that companies for completely other purposes are de facto, either implicitly or explicitly, collecting a huge amount of data, storage costs have plummeted and we discover every day new applications for things. If you are a responsible company you want to be careful but you also want to keep a lot of data because you may have a use for the historical track of the data and all this. The problem is if you think about the amount of data that is collected, take just geolocation on your phone, if you wanted in a pre-mobile phone era to have this data collected on the movements of let's say 65 million French people using a mobile phone you would have needed policemen to track on a daily basis and not on a little notepad where this person has gone and so on. This data would never have been accessible unless an explicit decision, were it to be made, to have someone to follow the movements of someone. The problem is that if you are a very well-meaning law enforcement agency, the existence of this trove of data is unbelievably telling because you know you can do a lot of good things with them. The problem is how to all make sure there is no abuse in the way it is used. So the challenge we have -- I love the fact that several of you are talking about words like "fuzzy," "blurred," "overlap"; the challenge we have is that the clear picture is more complex. It's multilayered, but also about shared responsibilities. And one of the challenges that I have here is that the mental framework within which the law enforcement agencies are in charge of security, the Civil Society actors are just in charge of protecting civil liberties. The businesses are just in charge of making money out of their activities. It's not so simple anymore. The fact is that as was said before in certain cases, voluntarily or under pressure companies are being now instrumentalized as law enforcement arms. Some of it is really bad. Some is useful because data is useful for law enforcement. When Civil Society actors fight, legitimately for protection of the privacy of individuals, they are also in the responsible role of making sure that what can be done to protect the individuals by a proper exploitation of the data is being done. Which leads me to the final point. I love the expression "Digital due process." The thing is we need new frameworks for the cooperation of actors. We cannot only -- and this is part of the response to the first question that was asked, on a personal basis and as the lessons we drove from the Internet and jurisdiction programme -it is an illusion to believe that everything will be solved by drafting documents, laws, treaties, whatever. This is not what it is about because in many cases you have a question of speed. You have a question of procedures of what is the appropriate level. The comment that was made by (indiscernible) regarding the judges or the courts system, there is the problem he mentioned but there is a problem of speed. In many cases, obtaining a full court decision takes a long time. So we are confronted with a problem. In some cases we want to be able to have a very quick action that is respectful of due process. And, on the other hand, if we respect the due process by getting always to courts either the courts are not completely aware of all the elements that have to be taken into account or they will take a long time to make a decision. So I personally would suggest as part of the process forward that one dimension that we are or have been talking about in this session is more or less how much can or should be accessed. There is an element which is what kind of procedures should be put in place, and here courts are important. But other mechanics of what we call here enhanced corporations are necessary between actors, between governments, Civil Society, private sector. But the most important element is that the national level is not sufficient. Because in many cases the platforms, particularly for cloud, are cross-border. If we wait until each national legislation elaborates its own service and its own provisions, in many cases it will not solve the cases where you really need to have access to data. So I would encourage the solutions to move in the direction of what kind of frameworks can be developed for cloud-based services, either for storage or for social media, where the cooperation between platforms, certain number of responsible governments willing to take the way and Civil Society actors that would have the capacity to monitor, for instance, logged requests would move forward. And this kind of framework for collaboration will actually be the topic of the workshop we have Thursday. But what is very important is we need to explore the range of tools, from complete court-mandated decisions in specific cases to very automated access with third-party control of the logging of the different actors. And within that you can have all modes of interfaces with the different actors. And I would like to throw something into the discussion, the notion of procedural interfaces. Platforms have procedures to implement their terms of service. The governments, law enforcement, and data protection authorities have their own procedures. But most of them are not sufficiently documented, they are not sufficiently transparent and they are not intra-operable. We need to work on that traceability of all those requests. >> Let me make three brief points. I agree absolutely responding to the lady's question from the start that surveillance policy-making needs to be much more multistakeholder. That's how you get the technological understanding Wendy was talking about, about the human rights input, input from privacy regulators who have to come in at the end to clear up a mess rather than contribute to the debate at the start. And technologists try very hard to explain these issue to policymakers, but often the voices of intelligence are very strong within government, politicians all from the left and right like to appear tough on crime, a favorite Tony Blair phrase, and I think that you absolutely need continued, meaningful, and strong oversight from the judiciary and from legislators. I'm not quite as comfortable as Bertrand about automating and examining, thinking transparency and multi-stakeholder auditing can go very far. Well, I would go much towards one end of the range. I think the U.S. has a number of things right on surveillance policy, one of which is Congress has much greater oversight of what the U.S. intelligence agencies do than almost any other nation, certainly compared to the UK. That is something legislators should look act. Secondly, on jurisdiction, I co-authored a report for the Global Network Initiative called "Digital Freedoms." One of the things we recommended in there was that when companies like Google, Facebook, RIM, are asked for voluntary or less voluntary cooperation by governments, and governments outside their main markets and certainly headquarters in the U.S., that by and large the route for law enforcement agencies in other countries should be through multilateral treaties, not by putting pressure. Google and Facebook do not want to be in the position of making judicial decisions. That's not their expertise. And the Council of Europe Cybercrime Convention is one framework that would meet some of those tests. I know some people are here, it's been strongly criticized and in some places it doesn't go far enough on human rights protections, but that could be worked on. That's an example of the framework that could deal with this much better than these voluntary data disclosures. Finally, on the more technological side, very interesting discussion between Bruce and Wendy and Mark about what causes people, users, to trust the systems or not trust. Absolutely -(Internet lost in conference room, captioner standing by) >> Hi. My name is Rutz, I'm from law enforcement. My question is for industry representatives here. My thing is about being visible and transparency. Do you have any regarding the crimes which the suspect is accused of? >> Classification is important and some are really important because we are trying to investigate them and if they that's going to be the case they will probably fly away. What's your opinion? >> RAPPORTEUR: To Marc, if someone else wants to reply from the panel, yes. >> MARC CRANDALL: Small tidbit of information -- by the way, thank you for the comment regarding law enforcement interaction -- before Google, many, many years ago, I was F.B.I. actually so I know the law enforcement concerns and I know the angle. But I also know after many other years how important it is to protect user information. Law enforcement does not necessarily provide what providers would otherwise want or maybe they don't want with regards to the basis for these legal requests. We may not know what these requests are, law enforcement may not want to share it with us. Furthermore, it is not necessarily within the provider's purview to make their own judgment call whether they should interact with law enforcement. If we are in the jurisdiction, that is subject to rule of law, then there should already be potentially some sort of review as to whether or not this is bad legal process. So we can't be the judge of whether or not process is accurate and that's assuming the law enforcement wants to share that data with us. But if law enforcement conducts themselves within the law and pursuant to guidelines outlined by the legislators, then it's much easier for providers to enact -from the policy perspective, continuing to engage regulators regarding clarity in law enforcement processes, is very very important, parity between online and offline and from a practical perspective where users need to develop trust, they should review what information is available regarding transparency so they can make their own risk assessment. Google's own transparency report is publicly available. We list how many requests we get from every country, what percentage we comply with. This is good information but we do that in a broad way so we don't jeopardize any specific investigation because we have to strike a balance. >> RAPPORTEUR: Bruce, we can have you next. >> BRUCE SCHNEIER: I think I want to answer the question in the back. A lot of metaphors are out of date. Friends are not what they were 100 years ago, a lot of the words we're using for these new things are old words but they're different. And that's a fundamental problem with communicating what we're doing to non-technologists. A waste basket on the computer is not a waste basket. It's a different thing. To speak to this person, I believe this is a pretty optimistic panel. Security people sound more pessimistic than they are because we deal in exceptions and the bad guys and bad actors. I feel very optimistic and that you're right, moving to the cloud and is beneficial in security for most people. My mother is much more secure because her data is on Gmail than her computer. She can't lose it. I don't have to rescue it. It's wonderful. For a lot of people that's true. That is why the cloud is so compelling. Everyone loves it when they lose their phone they get a new phone, push button and their contacts reappear by magic! We really like that. So don't take what we're talking about as overall pessimism. We are looking at the edges but the reason these things are happening is because they are so beneficial. >> RAPPORTEUR: We are going to give one minute to each participant for a closing but one minute, please, because we already run out of time. >> I think I have used my minute. >> BERTRAND LaCHAPELLE: I love the distinction Ian made between law enforcement and intelligence and surveillance. These two are different categories. He uses the word data minimization. He's absolutely right regarding data that platforms collect voluntarily from the users. But you cannot have that with the amount of personal data that users are putting on social network platforms which basically explain everything they have done from what they ate at breakfast and the rest. We are not talking about the same type of data. One is privacy and the other I call intimacy data. Finally, the lesson that we get from this environment is that we should try to move away mentally from the sharp distinctions of frontiers that separate jurisdiction A and B. You need not only have between different governments and agencies but also different operators and using in that regard a set of tools, one interesting trend I see if companies are beginning to hire law enforcement, former law enforcement officials, and Civil Society activists, actually, I think governments should consider more hiring people who have a previous corporate experience and Civil Society actors which is actually happening and it is a very good thing. Then there's a better understanding. >> Just 10 seconds. I agree with almost everything Bertrand said. I think that on the information users post about themselves, of course, there's only so far you can go with protecting privacy. You can't be too paternalistic. >> I want to thank you for the comment about not being a negative or harmful thing in the cloud and perhaps as a takeaway, as critics, we should always try to be very positive because they create a much more -- a dialogue you can work with. Instead of Civil Society constantly critiquing the government, instead, a have a positive critique on how we can all work together to create better solutions. So thank you for that. >> MARC CRANDALL: I'll forgo my statement. I have said enough. >> I'll say one more sentence. An important lesson is any laws and regulations need to be technologically invariant. The more we do that, the better we fare. If we do deal-specific technologies, they fail as soon as technologies change but these are fundamentally human interactions. If you focus on the human interaction, it doesn't matter how it happens or where it happens. Just a better way to look at this. >> I think we would also be closer to bridging the gap between these difficulties. I'm just promoting convention on data protection and cybercrime convention. >> The principles that we have are technologically invariant and the laws we adopt to meet them of due process and transparency and minimization should also similarly be broadly written and then applied as specific to the technology. >> RAPPORTEUR: Thank you, everyone. I'm sorry, but we have run out of time. Thank you, everyone, for coming. (Applause) (Session concluded) This text is being provided in a rough draft format. Communication Access Realtime Translation (CART) is provided in order to facilitate communication accessibility and may not be a totally verbatim record of the proceedings.