Design for Privacy September 2009 CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 1 Outline Engineering privacy Design of privacy tools Design for privacy in everyday software Obtaining informed consent CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 2 Engineering privacy CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 3 How Privacy Rights are Protected By policy By architecture – Protection through laws and organizational privacy policies – Must be enforced – Often requires mechanisms to obtain and record consent – Transparency facilitates choice and accountability – Technology facilitates compliance and reduces the need to rely solely on trust and external enforcement – Technology reduces or eliminates any form of manual processing or intervention by humans – Violations still possible due to bad actors, mistakes, government mandates – Protection through technology – Reduces the need to rely on trust and external enforcement – Violations only possible if technology fails or the availability of new data or technology defeats protections – Often viewed as too expensive or restrictive • Limits the amount of data available for data mining, R&D, targeting, other business purposes • May require more complicated system architecture, expensive cryptographic operations • Pay now or pay later CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 4 Privacy stages 0 identifiability identified Approach to privacy protection privacy by policy (notice and choice) 1 Linkability of data to personal identifiers linked • unique identifiers across databases • contact information stored with profile information linkable with reasonable & automatable effort • no unique identifies across databases • common attributes across databases • contact information stored separately from profile or transaction information not linkable with reasonable effort • no unique identifiers across databases • no common attributes across databases • random identifiers • contact information stored separately from profile or transaction information • collection of long term person characteristics on a low level of granularity • technically enforced deletion of profile details at regular intervals unlinkable • no collection of contact information • no collection of long term person characteristics • k-anonymity with large value of k pseudonymous 2 privacy by architecture 3 anonymous System Characteristics CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ Sarah Spiekermann and Lorrie Faith Cranor. Engineering Privacy. IEEE Transactions on Software Engineering. Vo. 35, No. 1, January/February, 2009, pp. 67-82. http://ssrn.com/abstract=1085333 Degrees of Identifiability 5 Design of Privacy Tools CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 6 Privacy tool examples Cookie managers Anonymizers Encryption tools Disk wiping utilities P3P user agents CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 7 Laptop Compubody Sock for privacy, warmth, and concentration in public spaces Created by Becky Stern http://sternlab.org/2008/04/body-technology-interfaces/ CIPP/IT Section Three | Privacy Protection Mechanisms CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 8 Issues to consider Privacy is a secondary task – Users of privacy tools often seek out these tools due to their awareness of or concern about privacy – Even so, users still want to focus on their primary tasks Users have differing privacy concerns and needs – One-size-fits-all interface may not work Most users are not privacy experts – Difficult to explain current privacy state or future privacy implications – Difficult to explain privacy options to them – Difficult to capture privacy needs/preferences Many privacy tools reduce application performance, functionality, or convenience CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 9 Case study: Tor Internet anonymity system Allows users to send messages that cannot be traced back to them (web browsing, chat, p2p, etc.) UI was mostly command line interface until recently 2005 Tor GUI competition – CUPS team won phase 1 with design for Foxtor! CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 10 One-size-doesn’t-fit-all problem Tor is configurable and different users will want to configure it in different ways – But most users won’t understand configuration options – Give users choices, not dilemmas We began by trying to understand our users – No budget, little time, limited access to users – So we brainstormed about their needs, tried to imagine them, and develop personas for them This process led to realization that our users had 3 categories of privacy needs – Basic, selective, critical Instead of asking users to figure out complicated settings, most of our configuration involves figuring out which types of privacy needs they have CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 11 CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 12 Understand primary task Anonymity is not a primary task What are the primary tasks our users are engaged in when they want anonymity? Lots of them …. Web browsing, chatting, file sharing, etc., but we speculate that browsing will be most frequent for most users So, instead of building anonymity tool that you can use to anonymize web browsing… … build a web browser with built in anonymity functions CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 13 Metaphors Because of performance issues and problems accessing some web sites through Tor, some users will want to turn the anonymity function on and off Important to make it easy for users to determine current state Communicate through visual symbol and readily understandable metaphor Brainstormed possibilities: torized/untorized, private/exposed, cloaked/uncloaked, masked/unmasked CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 14 CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 15 Design for privacy in every day software CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 16 Examples Ecommerce personalization systems – Concerns about use of user profiles Software that “phones home” to fetch software updates or refresh content, report bugs, relay usage data, verify authorization keys, etc. – Concerns that software will track and profile users Communications software (email, IM, chat) – Concerns about traffic monitoring, eavesdroppers Presence systems (buddy lists, shared spaces, friend finders) – Concerns about limiting when info is shared and with whom CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 17 Issues to consider Similar to issues to consider for privacy tools PLUS Users may not be aware of privacy issues up front – When they find out about privacy issues they may be angry or confused, especially if they view notice as inadequate or defaults as unreasonable Users may have to give up functionality or convenience, or spend more time configuring system for better privacy Failure to address privacy issues adequately may lead to bad press and legal action CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 18 The Prada NYC dressing room http://www.sggprivalite .com/ What aspects seem privacy invasive? How could the design be changed to reduce privacy concerns? CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 19 Amazon.com privacy makeover CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 20 Streamline menu navigation for customization Provide way to set up default rules Every time a user makes a new purchase that they want to rate or exclude they have to edit profile info – There should be a way to set up default rules • • • • Exclude all purchases Exclude all purchases shipped to my work address Exclude all movie purchases Exclude all purchases I had gift wrapped CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 22 Remove excluded purchases from profile Users should be able to remove items from profile If purchase records are needed for legal reasons, users should be able to request that they not be accessible online CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 23 Better: options for controlling recent history CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 24 Use personae Amazon already allows users to store multiple credit cards and addresses Why not allow users to create personae linked to each with option of keeping recommendations and history separate (would allow easy way to separate work/home/gift personae)? CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 25 Allow users to access all privacy-related options in one place Currently privacy-related options are found with relevant features Users have to be aware of features to find the options Put them all in one place But also leave them with relevant features CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 26 I didn’t buy it for myself How about an “I didn’t buy it for myself” check-off box (perhaps automatically checked if gift wrapping is requested) I didn’t buy it for myself CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 27 Other ideas for improving Amazon privacy interface? CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 28 Obtaining informed consent Many software products contain phone home features, for example, for performing software updates or monitoring usage patterns. In some cases software phones homes quite frequently, for example, to update phishing black lists or check for fresh image files. Users may be concerned that the software company is using these features to track or profile them. Thus it is important that the software is up front about the fact that it is phoning home. Furthermore, some users may wish to disable such features or be prompted every time before they phone home (due to privacy or other concerns), whereas other users are happy to have them operate automatically. Discuss the various approaches you have seen different software manufacturers take to addressing this problem. What do you like/dislike about them? How should phone home features be designed so that they facilitate informed consent? Describe an example user interface design and general principles that might be applied to specific cases. What sort of user studies should be performed to test this user interface design? CyLab Usable Privacy and Security Laboratory http://cups.cs.cmu.edu/ 29