New Approaches to Information Privacy – Revisiting the Purpose Limitation Principle

This was published in Digital Policy Portal on July 13, 2016.

Introduction

Last year, Mukul Rohatgi, the Attorney General of India, called into question existing jurisprudence of the last 50 years on the constitutional validity of the right to privacy.¹ Mohatgi was rebutting the arguments on privacy made against Aadhaar, the unique identity project initiated and implemented in the country without any legislative mandate.² The question of the right to privacy becomes all the more relevant in the context of events over the last few years—among them, the significant rise in data collection by the state through various e-governance schemes,³ systematic access to personal data by various wings of the state through a host of surveillance and law enforcement initiatives launched in the last decade,⁴ the multifold increase in the number of Indians online, and the ubiquitous collection of personal data by private parties.⁵

These developments have led to a call for a comprehensive privacy legislation in India and the adoption of the National Privacy Principles as laid down by the Expert Committee led by Justice AP Shah.⁶ There are privacy-protection legislation currently in place such as the Information Technology Act, 2000 (IT Act), which was enacted to govern digital content and communication and provide legal recognition to electronic transactions. This legislation has provisions that can safeguard—and dilute—online privacy. At the heart of the data protection provisions in the IT Act lies section 43A and the rules framed under it, i.e., Reasonable security practices and procedures and sensitive personal data information.⁷Section 43A mandates that body corporates who receive, possess, store, deal, or handle any personal data to implement and maintain ‘reasonable security practices’, failing which, they are held liable to compensate those affected. Rules drafted under this provision also mandated a number of data protection obligations on corporations such the need to seek consent before collection, specifying the purposes of data collection, and restricting the use of data to such purposes only. There have been questions raised about the validity of the Section 43A Rules as they seek to do much more than mandate in the parent provisions, Section 43A— requiring entities to maintain reasonable security practices.

Privacy as control?

Even setting aside the issue of legal validity, the kind of data protection framework envisioned by Section 43A rules is proving to be outdated in the context of how data is now being collected and processed. The focus of Section 43 A Rules—as well as that of draft privacy legislations in India⁸—is based on the idea of individual control. Most apt is Alan Westin’s definition of privacy: “the claim of individuals, groups, or institutions to determine for themselves when, how, and to what extent information about them is communicated to other.”⁹ Westin and his followers rely on the normative idea of “informational self- determination”, the notion of a pure, disembodied, and atomistic self, capable of making rational and isolated choices in order to assert complete control over personal information. More and more this has proved to be a fiction especially in a networked society.

Much before the need for governance of information technologies had reached a critical mass in India, Western countries were already dealing with the implications of the use of these technologies on personal data. In 1973, the US Department of Health, Education and Welfare appointed a committee to address this issue, leading to a report called ‘Records, Computers and Rights of Citizens.’¹⁰ The Committee’s mandate was to “explore the impact of computers on record keeping about individuals and, in addition, to inquire into, and make recommendations regarding, the use of the Social Security number.” The Report articulated five principles which were to be the basis of fair information practices: transparency; use limitation; access and correction; data quality; and security. Building upon these principles, the Committee of Ministers of the Organization for Economic Cooperation and Development (OECD) arrived at the Guidelines on the Protection of Privacy and Transborder Flows of Personal Data in 1980.¹¹ These principles— Collection Limitation, Data Quality, Purpose Specification, Use Limitation, Security Safeguards, Openness, Individual Participation and Accountability—are what inform most data protection regulations today including the APEC Framework, the EU Data Protection Directive, and the Section 43A Rules and Justice AP Shah Principles in India.

Fred Cate describes the import of these privacy regimes as such:

“All of these data protection instruments reflect the same approach: tell individuals what data you wish to collect or use, give them a choice, grant them access, secure those data with appropriate technologies and procedures, and be subject to third-party enforcement if you fail to comply with these requirements or individuals’ expressed preferences”¹²

This is in line with Alan Westin’s idea of privacy exercised through individual control. Therefore the focus of these principles is on empowering the individuals to exercise choice, but not on protecting individuals from harmful or unnecessary practices of data collection and processing. The author of this article has earlier written¹³ about the sheer inefficacy of this framework which places the responsibility on individuals. Other scholars like Daniel Solove,¹⁴ Jonathan Obar¹⁵ and Fred Cate¹⁶ have also written about the failure of traditional data protection practices of notice and consent. While these essays dealt with the privacy principles of choice and informed consent, this paper will focus on the principles of purpose limitation.

Purpose Limitation and Impact of Big Data

The principles of purpose limitation or purpose specification seeks to ensure the following four objectives:

1. Personal information collected and processed should be adequate and relevant to the purposes for which they are processed.
2. The entities collect, process, disclose, make available, or otherwise use personal information only for the stated purposes.
3. In case of change in purpose, the data’s subject needs to be informed and their consent has to be obtained.
4. After personal information has been used in accordance with the identified purpose, it has to be destroyed as per the identified procedures.

The purpose limitation along with the data minimisation principle—which requires that no more data may be processed than is necessary for the stated purpose—aim to limit the use of data to what is agreed to by the data subject. These principles are in direct conflict with new technology which relies on ubiquitous collection and indiscriminate uses of data. The main import of Big Data technologies on the inherent value in data which can be harvested not by the primary purposes of data collection but through various secondary purposes which involve processing of the data repeatedly.¹⁷Further, instead to destroying the data when its purpose has been achieved, the intent is to retain as much data as possible for secondary uses. Importantly, as these secondary uses are of an inherently unanticipated nature, it becomes impossible to account for it at the stage of collection and providing the choice to the data subject.

Followers of the discourse on Big Data would be well aware of its potential impacts on privacy. De-identification techniques to protect the identities of individuals in dataset face a threat from an increase in the amount of data available either publicly or otherwise to a party seeking to reverse-engineer an anonymised dataset to re-identify individuals. ¹⁸ Further, Big Data analytics promise to find patterns and connections that can contribute to the knowledge available to the public to make decisions. What is also likely is that it will lead to revealing insights about people that they would have preferred to keep private.¹⁹In turn, as people become more aware of being constantly profiled by their actions, they will self-regulate and ‘discipline’ their behaviour. This can lead to a chilling effect.²⁰ Meanwhile, Big Data is also fuelling an industry that incentivises businesses to collect more data, as it has a high and growing monetary value. However, Big Data also promises a completely new kind of knowledge that can prove to be revolutionary in fields as diverse as medicine, disaster-management, governance, agriculture, transport, service delivery, and decision-making.²¹ As long as there is a sufficiently large and diverse amount of data, there could be invaluable insights locked in it, accessing which can provide solutions to a number of problems. In light of this, it is important to consider what kind of regulatory framework is most suitable which could facilitate some of the promised benefits of Big Data and at the same time mitigate its potential harm. This, coupled with the fact that the existing data protection principles have, by most accounts, run their course, makes the examination of alternative frameworks even more important. This article will examine some alternate proposals made to the existing framework of purpose limitation below.

Harms-based approach

Some scholars like Fred Cate²² and Daniel Solove²³ have argued that there is a need for the primary focus of data protection law to move from control at the stage of data collection to actual use cases. In his article on the failure of Fair Information Practice Principles,²⁴Cate puts forth a proposal for ‘Consumer Privacy Protection Principles.’ Cate envisions a more interventionist role of the data protection authorities by regulating information flows when required, in order to protect individuals from risky or harmful uses of information. Cate’s attempt is to extend the principles of consumer protection law of prevention and remedy of harms.

In a re-examination of the OECD Privacy Principles, Cate and Viktor Mayer Schöemberger attempt to discard the use of personal data to only purposes specified. They felt that restricting the use of personal to only specified purposes could significantly threaten various research and beneficial uses of Big Data. Instead of articulating a positive obligations of what personal data collected could be used for, they attempt to arrive at a negative obligation of use-cases prevented by law. Their working definition of the Use specification principle broaden the scope of use cases by only preventing use of data “if the use is fraudulent, unlawful, deceptive or discriminatory; society has deemed the use inappropriate through a standard of unfairness; the use is likely to cause unjustified harm to the individual; or the use is over the well-founded objection of the individual, unless necessary to serve an over-riding public interest, or unless required by law.”²⁵

While most standards in the above definition have established understanding in jurisprudence, the concept of unjustifiable harm is what we are interested in. Any theory of harms-based approach goes back to John Stuart Mill’s dictum that the only justifiable purpose to exert power over the will of an individual is to prevent harm to others. Therefore, any regulation that seeks to control or prevent autonomy of individuals (in this case, the ability of individuals to allow data collectors to use their personal data, and the ability of data collectors to do so, without any limitation) must clearly demonstrate the harm to the individuals in question.

Fred Cate articulates the following steps to identify tangible harm and respond to its presence:²⁶

1. Focus on Use — Actual use of the data should be considered, not mere possession. The assumption is that the collection, possession, or transfer of information do not significantly harm people, rather it is the use of information following such collection, possession, or transfer.
2. Proportionality — Any regulatory measure must be proportional to the likelihood and severity of the harm identified.
3. Per se Harmful Uses — Uses which are always harmful must be prohibited by law
4. Per se not Harmful Uses — If uses can be considered inherently not harmful, they should not be regulated.
5. Sensitive Uses — In case where the uses are not per se harmful or not harmful, individual consent must be sought for using that data for those purposes.

The proposal by Cate argues for what is called a ‘use based system’, which is extremely popular with American scholars. Under this system, data collection itself is not subject to restrictions; rather, only the use of data is regulated. This argument has great appeal for both businesses who can reduce their overheads significantly if consent obligations are done away with as long as they use the data in ways which are not harmful, as well as critics of the current data protection framework which relies on informed consent. Lokke Moerel explains the philosophy of ‘harms based approach’ or ‘use based system’ in United States by juxtaposing it against the ‘rights based approach’ in Europe.²⁷ In Europe, rights of individuals with regard to processing of their personal data is a fundamental human right and therefore, a precautionary principle is followed with much greater top-down control upon data collection. However, in the United States, there is a far greater reliance on market mechanisms and self-regulating organisations to check inappropriate processing activities, and government intervention is limited to cases where a clear harm is demonstrable.²⁸

Continuing research by the Centre for Information Policy Leadership under its Privacy Risk Framework Project looks at a system of articulating what harms and risks arising from use of collected data. They have arrived a matrix of threats and harms. Threats are categorised as —a) inappropriate use of personal information and b) personal information in the wrong hands. More importantly for our purposes, harms are divided into: a) tangible harms which are physical or economic in nature (bodily harm, loss of liberty, damage to earning power and economic interests); b) intangible harms which can be demonstrated (chilling effects, reputational harm, detriment from surveillance, discrimination and intrusion into private life); and c) societal harm (damage to democratic institutions and loss of social trust).²⁹For any harms-based system, a matrix like above needs to emerge clearly so that regulation can focus on mitigating practices leading to the harms.

Legitimate interests

Lokke Moerel and Corien Prins, in their article “Privacy for Homo Digitalis – Proposal for a new regulatory framework for data protection in the light of Big Data and Internet of Things”³⁰ use the ideal of responsive regulation which considers empirically observable practices and institutions while determining the regulation and enforcement required. They state that current data protection frameworks—which rely on mandating some principles of how data has to be processed—is exercised through merely procedural notification and consent requirements. Further, Moerel and Prins feel that data protection law cannot only involve a consideration of individual interest but also needs to take into account collective interest. Therefore, the test must be a broader assessment than merely the purpose limitation articulating the interests of the parties directly involved, but whether a legitimate interest is achieved.

Legitimate interest has been put forth as an alternative to the purpose limitation. Legitimate is not a new concept and has been a part of the EU Data Protection Directive and also finds a place in the new General Data Protection Regulation. Article 7 (f) of the EU Directive³¹ provided for legitimate interest balanced against the interests or fundamental rights and freedoms of the data subject as the last justifiable reason for use of data. Due to confusion in its interpretation, the Article 29 Working Party, in 2014,³²looked into the role of legitimate interest and arrived at the following factors to determine the presence of a legitimate interest— a) the status of the individual (employee, consumer, patient) and the controller (employer, company in a dominant position, healthcare service); b) the circumstances surrounding the data processing (contract relationship of data subject and processor); c) the legitimate expectations of the individual.

Federico Ferretti has criticised the legitimate interest principle as vague and ambiguous. The balancing of legitimate interest in using the data against fundamental rights and freedoms of the data subject gives the data controllers some degree of flexibility in determining whether data may be processed; however, this also reduces the legal certainty that data subject have of their data not being used for purposes they have not agreed to.³³However, it is this paper’s contention that it is not the intent of the legitimate interest criteria but the lack of consensus on its application which creates an ambiguity. Moerel and Prins articulate a test for using legitimate interest which is cognizant of the need to use data for the purpose of Big Data processing, as well as ensuring that the rights of data subjects are not harmed.

As demonstrated earlier, the processing of data and its underlying purposes have become exceedingly complex and the conventional tool to describe these processes ‘privacy notices’ are too lengthy, too complex and too profuse in numbers to have any meaningful impact.³⁴The idea of information self-determination, as contemplated by Westin in American jurisprudence, is not achieved under the current framework. Moerel and Prins recommend five factors³⁵ as relevant in determining the legitimate interest. Of the five, the following three are relevant to the present discussion:

1. Collective Interest — A cost-benefit analysis should be conducted, which examines the implications for privacy for the data subjects as well as the society, as a whole.
2. The nature of the data — Rather than having specific categories of data, the nature of data needs to be assessed contextually to determine legitimate interest.
3. Contractual relationship and consent not independent grounds — This test has two parts. First, in case of contractual relationship between data subject and data controller: the more specific the contractual relationship, the more restrictions apply to the use of the data. Second, consent does not function as a separate principle which, once satisfied, need not be revisited. The nature of the consent (opportunities made available to data subject, opt in/opt out, and others) will continue to play a role in determining legitimate interest.

Conclusion

Replacing the purpose limitation principles with a use-based system as articulated above poses the danger of allowing governments and the private sector to carry out indiscriminate data collection under the blanket guise that any and all data may be of some use in the future. The harms-based approach has many merits and there is a stark need for more use of risk assessments techniques and privacy impact assessments in data governance. However, it is important that it merely adds to the existing controls imposed at data collection, and not replace them in their entirety. On the other hand, the legitimate interests principle, especially as put forth by Moerel and Prins, is more cognizant of the different factors at play — the inefficacy of existing purpose limitation principles, the need for businesses to use data for purposes unidentified at the stage of collection, and the need to ensure that it is not misused for indiscriminate collection and purposes. However, it also poses a much heavier burden on data controllers to take into account various factors before determining legitimate interest. If legitimate interest has to emerge as a realistic alternative to purpose limitation, there needs to be greater clarity on how data controllers must apply this principle.

Endnotes

1. Prachi Shrivastava, “Privacy not a fundamental right, argues Mukul Rohatgi for Govt as Govt affidavit says otherwise,” Legally India, Jyly 23, 2015, http://www.legallyindia.com/Constitutional-law/privacy-not-a-fundamental-right-argues-mukul-rohatgi-for-govt-as-govt-affidavit-says-otherwise.
2.  Rebecca Bowe, “Growing Mistrust of India’s Biometric ID Scheme,” Electronic Frontier Foundation, May 4, 2012, https://www.eff.org/deeplinks/2012/05/growing-mistrust-india-biometric-id-scheme.
3. Lisa Hayes, “Digital India’s Impact on Privacy: Aadhaar numbers, biometrics, and more,” Centre for Democracy and Technology, January 20, 2015, https://cdt.org/blog/digital-indias-impact-on-privacy-aadhaar-numbers-biometrics-and-more/.
4. “India’s Surveillance State,” Software Freedom Law Centre, http://sflc.in/indias-surveillance-state-our-report-on-communications-surveillance-in-india/.
5. “Internet Privacy in India,” Centre for Internet and Society, http://cis-india.org/telecom/knowledge-repository-on-internet-access/internet-privacy-in-india.
6. Vivek Pai, “Indian Government says it is still drafting privacy law, but doesn’t give timelines,” Medianama, May 4, 2016, http://www.medianama.com/2016/05/223-government-privacy-draft-policy/.
7. Information Technology (Intermediaries Guidelines) Rules, 2011,
   http://deity.gov.in/sites/upload_files/dit/files/GSR314E_10511%281%29.pdf.
8. Discussion Points for the Meeting to be taken by Home Secretary at 2:30 pm on 7-10-11 to discuss the drat Privacy Bill, http://cis-india.org/internet-governance/draft-bill-on-right-to-privacy.
9. Alan Westin, Privacy and Freedom (New York: Atheneum, 2015).
10. US Secretary’s Advisory Committee on Automated Personal Data Systems, Records, Computers and the Rights of Citizens, http://www.justice.gov/opcl/docs/rec-com-rights.pdf.
11. OECD Guidelines on the Protection of Privacy and Transborder Flows of Personal Data, http://www.oecd.org/sti/ieconomy/oecdguidelinesontheprotectionofprivacyandtransborderflowsofpersonaldata.htm
12. Fred Cate, “The Failure of Information Practice Principles,” in Consumer Protection in the Age of the Information Economy, ed. Jane K. Winn (Burlington: Aldershot, Hants, England, 2006) http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1156972.
13. Amber Sinha and Scott Mason, “A Critique of Consent in Informational Privacy,” Centre for Internet and Society, January 11, 2016, http://cis-india.org/internet-governance/blog/a-critique-of-consent-in-information-privacy.
14. Daniel Solove, “Privacy self-management and consent dilemma,” Harvard Law Review 126, (2013): 1880.
15. Jonathan Obar, “Big Data and the Phantom Public: Walter Lippmann and the fallacy of data privacy self management,” Big Data and Society 2(2), (2015), doi: 10.1177/2053951715608876.
16. Supra Note 12.
17. Supra Note 14.
18. Paul Ohm, “Broken Promises of Privacy: Responding to the Surprising Failure of Anonymization” available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1450006; Arvind Narayanan and Vitaly Shmatikov, “Robust De-anonymization of Large Sparse Datasets” available at https://www.cs.utexas.edu/~shmat/shmat_oak08netflix.pdf.
19. D. Hirsch, “That’s Unfair! Or is it? Big Data, Discrimination and the FTC’s Unfairness Authority,” Kentucky Law Journal, Vol. 103, available at: http://www.kentuckylawjournal.org/wp-content/uploads/2015/02/103KyLJ345.pdf
20. A Marthews and C Tucker, “Government Surveillance and Internet Search Behavior”, available at http://ssrn.com/abstract=2412564; Danah Boyd and Kate Crawford, “Critical Questions for Big Data: Provocations for a cultural, technological, and scholarly phenomenon”, Information, Communication & Society, Vol. 15, Issue 5, (2012).
21. Scott Mason, “Benefits and Harms of Big Data”, Centre for Internet and Society, available at http://cis-india.org/internet-governance/blog/benefits-and-harms-of-big-data#_ftn37.
22. Cate, “The Failure of Information Practice Principles.”
23. Solove, “Privacy self-management and consent dilemma,” 1882.
24. Cate, “The Failure of Information Practice Principles.”
25. Fred Cate and Viktor Schoenberger, “Notice and Consent in a world of Big Data,” International Data Privacy Law 3(2), (2013): 69.
26. Solove, “Privacy self-management and consent dilemma,” 1883.
27. Lokke Moerel, “Netherlands: Big Data Protection: How To Make The Draft EU Regulation On Data Protection Future Proof”, Mondaq, March 11. 2014, http://www.mondaq.com/x/298416/data+protection/Big+Data+Protection+How+To+Make+The+Dra%20ft+EU+Regulation+On+Data+Protection+Future+Proof%20al%20Lecture.
28. Moerel, “Netherlands: Big Data Protection.”
29. Centre for Information Policy Leadership, “A Risk-based Approach to Privacy: Improving Effectiveness in Practice,” Hunton and Williams LLP, June 19, 2014, https://www.informationpolicycentre.com/uploads/5/7/1/0/57104281/white_paper_1-a_risk_based_approach_to_privacy_improving_effectiveness_in_practice.pdf.
30. Lokke Moerel and Corien Prins, “Privacy for Homo Digitalis: Proposal for a new regulatory framework for data protection in the light of Big Data and Internet of Things”, Social Science Research Network, May 25, 2016, http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2784123.
31. EU Directive 95/46/EC – The Data Protection Directive, https://www.dataprotection.ie/docs/EU-Directive-95-46-EC-Chapter-2/93.htm.
32. Article 29 Data Protection Working Party, “Opinion 06/2014 on the notion of legitimate interests of the data controller under Article 7 of Directive 95/46/EC,” http://ec.europa.eu/justice/data-protection/article-29/documentation/opinion-recommendation/files/2014/wp217_en.pdf.
33. Frederico Ferretti, “Data protection and the legitimate interest of data controllers: Much ado about nothing or the winter of rights?,” Common Market Law Review 51(2014): 1-26. http://bura.brunel.ac.uk/bitstream/2438/9724/1/Fulltext.pdf.
34. Sinha and Mason, “A Critique of Consent in Informational Privacy.”
35. Moerel and Prins, “Privacy for Homo Digitalis.”