Will European privacy regulators raise the bar for government cloud computing?

Jeff Gould by Jeff Gould, SafeGov.org
Friday, September 21, 2012

For those who follow government computing trends, the biggest story of 2012 in the U.S. has been the accelerating adoption of cloud services by Federal agencies as well as by State and Local governments. This growth has been fostered in large part by the admirably proactive stance in favor of cloud taken by the White House’s Office of Management and Budget (OMB). It has also been propelled by the FedRAMP program[1], which streamlines the procedures used to vet the security features of commercial cloud solutions. At SafeGov we enthusiastically endorse this trend and look forward to the cost savings and improvements in citizen services it will bring to all levels of government.

But while U.S. government use of cloud services surges, U.S. regulators have paid relatively little attention to the emerging issue of data confidentiality in the cloud. The focus of Federal cloud standardization efforts such as the NIST requirements[2] built into FedRAMP has been data security, not privacy and confidentiality. In Europe, however, the picture looks very different. The European public sector is approaching the cloud with caution: governments are keenly interested in the potential benefits, but have not yet issued the kind of top-down mandate for rapid migration that we’ve seen in the U.S. At the same time, European regulators are much further along the path toward a modernized regulatory regime for cloud computing.

The key development looming on the horizon in Europe is the proposed new EU Data Protection Regulation. This draft legislation represents a sweeping revision of the 1995 EU Privacy Directive and is currently the subject of intense scrutiny by interested stakeholders. Observers expect it to be passed by the EU Parliament sometime in early 2014. In the short term, the most significant event is the investigation of Google’s privacy policy that the French Data Protection Authority – the CNIL[3] – is conducting at the request of the Article 29 Working Party (the association of European DPAs).

Before we assess the impact of European privacy regulations on government cloud computing, let’s take a step back to review the recent debate over online privacy in the U.S. As visitors to this forum may know, SafeGov contributors were among the first to identify the mismatch between the privacy policies of giant web advertising companies like Google and the requirements of safe cloud computing in a government or educational environment. Recall that the new Google privacy policy introduced in March of this year allows the Mountain View firm to combine all of the vast knowledge it gleans from tracking a user’s activity across its many web services (Gmail, Docs, Search, YouTube, DoubleClick, etc.) into a single “master profile”. This profile can then be intensively data mined to select the most profitable ads to serve to that user. Scaled up to tens and even hundreds of millions of users, this profiling technique yields an extraordinarily profitable business model that has made Google the most successful advertising firm in history[4].

Privacy advocates on both sides of the Atlantic have objected to this business model on the grounds that web users are not informed that they are being tracked in this manner and are not given the opportunity to opt out. We also note that the European DPAs asked Google to delay implementation of the policy until it could be investigated, but Google declined. SafeGov itself does not take a position on business models deployed by consumer advertising firms. We recognize that opinions on this difficult and sensitive question will differ. Web advertising (which does not necessarily require hidden user tracking) can be a healthy form of technological innovation that offers significant benefits to consumers.

However, our contributing experts have pointed out on many occasions that the kind of stealthy user profiling and systematic data mining of user content that has become the norm on the consumer web is absolutely unacceptable when performed in cloud services provided under contract to governments or schools. I believe that our experts who have spoken out on this issue are on solid ground. Imagine for example that a cloud provider decided to apply the same data mining algorithms it uses for consumer ad targeting to the email traffic of tens or hundreds of thousands of government users or school children. Even if no personal information of individual users was disclosed to advertisers, the power of these algorithms to identify trending topics and keywords in user content could be of immense economic value. In the case of sensitive government information, it could also represent a grave threat to the security of nations. It is for these reasons that SafeGov has called on all cloud service providers to create separate privacy policies for public sector users that expressly ban these practices.

Now what of the European regulators? As noted above, the French DPA - the CNIL - was assigned the task last February of investigating Google’s new privacy policy in order to determine whether it complies with existing European data protection rules. The CNIL is expected to present its initial findings on the Google policy to its European peers sometime this month. It is important to understand that although the CNIL is a French institution, it is not acting on behalf of the French government, but on that of the association of European DPAs (the Article 29 Working Party). These DPAs are national regulatory bodies whose members are appointed by their national governments, but which operate as independent authorities (in much the same way that the FTC and the FCC do in the U.S.). Their mandate is to enforce European and national laws concerning data protection and online privacy.

While nothing has yet leaked to the press regarding the CNIL’s findings (the contrast on this point with the American FTC is noteworthy), past statements of the CNIL and the Article 29 Working Party allow us to anticipate the likely direction the regulator will take. First, it is highly probable that the CNIL will find that Google’s privacy policy indeed does not fully comply with European law. This much was already implied in the statements of Article 29 Working Party Chairman Jacob Kohnstamm last February and by the decision to entrust an investigation to the CNIL. Second, it is unlikely that the CNIL will adopt a punitive stance toward Google, for example by imposing a fine. European law gives the DPAs the power to fine companies that violate the rules, and several DPAs (including the CNIL) have already inflicted fines on Google for that firm’s conduct in the so-called Wi-Spy scandal. But in this case it is more likely that the CNIL and the other DPAs will politely ask Google to change its privacy policy in ways that make it compatible with European laws.

What changes might the Europeans seek in Google’s privacy policy? Any answer to this question before the release of the CNIL’s report is of course purely speculative. Certainly we can expect the regulator to require that Google do more to disclose to users the extent of its data gathering and to offer them more explicit opportunities to opt out. In recent months many web sites in Europe have begun to implement the new EU cookie rules that require increased disclosure and express consent prior to the serving of web cookies to user browsers. We might expect similar requirements to be imposed on Google and its web advertising peers (Facebook, Yahoo, Hotmail and Bing, etc.).

But at SafeGov our mission is government computing. We don’t know at this point whether the CNIL will express an opinion on the suitability of Google’s privacy policy for cloud services delivered to government customers. We note optimistically that the CNIL asked Google whether its new privacy policy applied to users of Google Apps for Education and Google Apps for Business (of which Google Apps for Government is a derivative)[5]. However, the regulator may prefer to focus its initial report on a broad outline of the changes it wishes to see in Google’s privacy policy, rather than drilling down to issues that confront specific sectors such as government or education.

In any case, observers can be confident that the debate on the topic of the confidentiality and safety of government data in the cloud is only just beginning. The CNIL’s findings on behalf of the Article 29 Working Party, whatever they are, will be only the first step in a long road. As Europe prepares a fundamental revision of its data protection and online privacy law, that road will ultimately lead to significant changes in the privacy practices and perhaps even in the business models of all web advertising firms that wish to do business in Europe. These changes will inevitably encompass the rules that govern the cloud services provided to European governments and schools. We hope that Europe’s Data Protection Authorities will recognize the need for dedicated privacy policies that guarantee users in these critical sectors of the European economy protection from the user profiling and data mining practices of the online consumer advertising industry.

Editor's note: This article also appeared on AOL Government.


[1] Federal Risk and Authorization Management Program.

[2] National Institute of Standards and Technology.

[3] Commission Nationale de l'Informatique et des Libertés (“National Commission on Computing and Liberties”).

[4] Google is not the only web firm to use this model. Facebook does essentially the same thing, although it has less raw data about users’ behavior outside of its own site. Even Microsoft has recently adopted a unified privacy policy for its consumer services such as the Bing search engine and the Hotmail ad-supported free email service (now known as Outlook.com) that may allow it to engage in practices similar to Google. However, a critical difference between the Google and Microsoft policies is the fact that Microsoft, unlike Google, has a specific privacy policy for enterprise and government users. This policy, which applies to users of the Office 365 service (equivalent to Google Apps), explicitly rules out data mining and user profiling: “Office 365 does not build advertising products out of Customer Data. We don’t scan your email or documents for building analytics, data mining, advertising, or improving the service.” SafeGov would like to see Google and other providers of cloud services to government and public sector users adopt similar language in their privacy policies.

[5] See Question 47 in the CNIL’s second questionnaire addressed to Google (dated May 22, 2012).

More information

Post a comment

Sign in to comment.

Not yet registered? Join the debate