Privacy Accountability and Enforceability Jamie Yoo April 11

Privacy: Accountability and Enforceability Jamie Yoo April 11, 2006 CPSC 457: Sensitive Information in a Wired World

Control of Personal Information ® Basic Problem: ® Data subject lacks control of sensitive information after initial disclosure ® Organizations lack control of the information that they manage once they disclose it to third parties

Fair Information Practices Principles ® Collection limitation ® Data quality ® Security safeguards ® Openness ® Purpose specification ® Use limitation ® Individual participation ® Accountability

Fair Information Practice Principles are guiding principles not law. Problem: Companies will claim to follow fair information practice principles but degree of implementation varies among companies.

Example: Data Resellers

Data Resellers (Brokers) ® Information Resellers are businesses that collect and aggregate personal information from multiple sources and make it available to their customers.

Privacy Problems Collection Limitation Information Resellers Generally Do Not Limit Data Collection to Specific Purposes and Do Notify Data Subjects

Collection Limitation Problem Resellers are limited only by laws that apply to specific kinds of information. ® Otherwise, resellers aggregate unrestricted amounts of personal information. ® No provisions are made to notify the data subjects when the reseller obtains personal data. ® ® ® Individuals are not afforded an opportunity to express or withhold their consent because many times resellers do not have a direct relationship with data subjects. Some offer an “opt-out” option but usually under limited circumstances for specific types of data and under specific conditions.

Privacy Problems Data Quality Information Resellers Do Not Ensure That Personal Information They Provide is Accurate for Specific Purposes

Data Quality Problem No standard mechanism for verifying the accuracy of the data obtained ® Some privacy policies state that resellers expect their data to contain some errors ® Varying policies regarding correction of data determined to be inaccurate as obtained by them ® ® ® Because they are not the original source of the personal information, information resellers generally direct individuals to the original sources to correct any errors. That is, data that may be perfectly adequate for one purpose may not be precise enough or appropriate for another purpose.

Privacy Problems Purpose Specification Information Resellers’ Specification of the Purpose of Data Collection Consists of Broad Descriptions of Business Categories

Purpose Specification Problem Information resellers specify purpose in a broad, general way by describing the types of businesses that use their data. ® They generally do not designate specific intended uses for each of their data collections. ® Generally, resellers obtain information that has already been collected for a specific purpose and make that information available to their customers, who in turn have a much broader variety of purposes for using it. ®

Privacy Problems Accountability Often times, data subjects do not even know that data resellers are selling their personal information, so accountability from an individual data subject’s standpoint is less than ideal.

Problems with Current “Solutions”

Limitations of Legislation ® Either too broad or too specific ® Slow to change ® Difficulty to enforce ® Especially across borders

Limitations of the FTC ® The Commission prosecutes “unfair and deceptive practices” violations. ® However, usually “letters from consumers or businesses, Congressional inquiries, or articles on consumer or economic subjects” triggers an FTC investigation. ® Unfortunately, data subjects are often not even aware of privacy violations, especially since they are not usually aware of specific instances of data disclosures by authorized data recipients to third parties

P 3 P ® ® ® P 3 P is a semi-structured privacy policy specification language that allows an organization to specify its website privacy practices in a machine-readable format. A P 3 P policy expresses the privacy practices related to the particular page or pages it governs; it covers any information collection on those pages, the purposes of that collection, the information recipient, and the length of that information’s retention. Specifications are checked by a browser/user agent, against user-specified preferences, to determine whether the organization follows user-acceptable privacy practices. User’s agent allows the load of a page, prevents the load, or notifies the user that the site does not (or may not) comply with the user’s preset preferences. Limitations: After initial disclosure of personal information, user has no mechanism for enforcement.

Enterprise Privacy Authorization Language (EPAL) Interoperability language for exchanging privacy policy in a structured format between applications/enterprises ® Access-centric ® Based on “strong associations” of fine-grained privacy policies (“sticky policies”) ® EPAL Policy: Defines lists of hierarchies of ® ® ® ® Data categories User categories Purposes Actions Obligations Conditions

Example of EPAL Rule Privacy Policy (informal): Allow a sales agent or a sales supervisor to collect a customer's data for order entry if the customer is older than 13 years of age and the customer has been notified of the privacy policy. Delete the data 3 years from now. EPAL Privacy Rule: ruling allow user category sales department action store data category customer-record purpose order-processing condition the customer is older than 13 years of age obligation delete the data 3 years from now

Current Usage Scenario Service Provider P 3 P Policy Consumer Transmits User Accepts or Rejects Agent Configures Reveals Personal Information Respects EPAL Policy Consumer bases her decision on announced P 3 P policy, which is not formally related to operative EPAL policy.

Issues ® Privacy promises made without mechanism for enforcement The “stickiness” of policies is not enforceable ® Too much trust in the enterprise ® Leakages can still happen ® ® Minimal user involvement (negotiation) ® Privacy management is more than authorization

Recommendation

Third Party Auditor: Tracing & Auditing Data ® Trusted third party to provide a mechanism for auditing/logging each disclosure Manages and records release of data (encryption) ® Validates privacy policy adhering environment of recipient ® Creates a paper trail ® ® Legislation to prosecute privacy violations ® In particular, legislation regulating the data brokering industry (ex: require deletion/renewal of data after x years, etc) ® Auditing should help with prosecution

Suggested Scenario Personal Data (encrypted) Privacy Policies (EPAL rules) Enterprise 1 Data Subject Enterprise 2 Decryption Key Trust Auditing and Tracing Authority

Details ® Identity-Based Encryption: Data Sender encrypts data package (data + privacy policy), Trusted Auditing Authority provides decryption keys to verified Data Recipient ® Trusted Computing defined by Auditor could be used to ensure privacy policy adhering environment ® Would allow for greater “stickiness” of policies to data (tamper-proof data tags): ® ® Privacy policy rules (ex: expiration date, etc) Digital signatures to indicate where the data came from (third party or directly from the user)

Limitations Difficult to build a trusted network of this type ® Inherent technical difficulty in representing privacy policies as machine-readable code remains ® Ex: A very large number of EPAL rules required to implement HIPAA, making it difficult to implement as well as maintain. ® Future of Trusted Computing is unknown ® Regardless of technical solutions, there must be legislative enforcement to encourage this type of rigorous auditing and also to prosecute violations ®