1. 18734: Foundations of Privacy
Course Review
Anupam Datta
CMU
Fall 2017

2. Personal Information is Everywhere

3. Privacy and Fairness Problems
Collection
Inference
Use
Collection: Massive online tracking, data brokers, govt surveillance
Inference: Facebook – sexual orientation from friendship graph, Target – pregnancy status from purchase history
Use: pricing, bias
Dissemination: search engines, RECAP and online court records
Dissemination

4. Organizing Questions What is privacy? What is fairness?
From philosophical and legal conceptions to computer science and engineering
Inspiration from conceptions, but greater precision often through greater specificity
How can we protect privacy and fairness?
Beyond creating laws and institutions
Computational mechanisms

5. An Organizing Viewpoint
Privacy as a right to restrictions on personal information flow
Collection
Inference
Use
Dissemination

6. Privacy enhancing technology adoption
Some drivers
User/citizen trust
Ethics/culture
Internal team
Regulation
Public opinion

7. Privacy Problems

8. Module I: Privacy through Accountability
Collection
Use
Dissemination

9. Web Privacy: Online Tracking
Collection 64
Independent tracking mechanisms on average on top-50 sites

10. Healthcare Privacy Privacy Expectations Hospital Analyst Patient
information
Patient
information
Patient
information
To understand why we want to formalize purpose restrictions, let’s start with an example. A patient goes to a hospital and provides medical information to a physician. The physician needs help and passed some of this information along to a nurse. The nurse then turns around and sells the information to a drug company that uses it for marketing. Since some patients object to such uses of their health information and we, as a society, want to encourage open communication between patients and their physicians, we have adopted privacy policies, such as HIPAA, that prohibit such uses of patient information without their consent. To ensure that employees comply with these policies, hospitals employ auditors who examine accesses to and transmissions of protected information looking for actions that violate the privacy policies in place.
Drug Company
Patient
Physician
Nurse

11. HIPAA Privacy Rule Use Dissemination
A covered entity may disclose an individual’s protected health information (phi) to law-enforcement officials for the purpose of identifying an individual if the individual made a statement admitting participating in a violent crime that the covered entity believes may have caused serious physical harm to the victim

12. Web Advertising Example privacy policies:
Use
Example privacy policies:
Not use detailed location (full IP address) for advertising
Not use health information for advertising

13. Privacy Compliance for Bing
Use
Setting:
Auditor has access to source code

14. Web Privacy: Advertising
Use
Ads
Sensitive
Information
(e.g., race, health
information)
Google
Confounding
Inputs

15. Module I: Privacy through Accountability
Formalize Privacy Policies
Precise semantics of privacy concepts
(restrictions on personal information flow)
Enforce Privacy Policies
Accountability
Detect
Explain
Correct
We have a body of work on understanding concepts in these privacy policies --- laws and enterprise policies --- at an operational level and on algorithms and computer systems to aid in the task of enforcing these policies.
Our enforcement regime comprises of audit and accountability mechanisms to detect policy violations, identify agents to blame for the violations, and impose appropriate sanctions to deter future violations.

16. Module I: Learning Outcomes
Understanding of real-world privacy policies and laws
Methods for detecting privacy violations
Experience with audit tools for healthcare privacy
Experience with web tracking investigation tool

17. Module II: Protecting Privacy and Fairness in Big Data Analytics
Collection
Inference
Use
Dissemination
CMU

18. Database Privacy Goals
Government,
marketers,
researchers, …
Health records
Census data
Web search records
Conflicting goals:
Provide useful information
Protect individual privacy 18

19. Inference
CMU

20. Inference
CMU

21. Privacy Solutions
Collection
Inference
Dissemination

22. Module II: Learning Outcomes
Understanding of pitfalls in anonymizing databases
Understanding of methods for releasing privacy- preserving statistics and their limitations
Understanding bias in machine learning and corrective measures
Understanding transparency (explanations) for decisions of machine learning systems
CMU

23. Module III: Cryptographic Mechanisms for Privacy Protection
Collection

24. Anonymous Communication

25. Digital Cash ... IN: scriptSig ... IN: scriptSig A OUT: OUT:
scriptPub A,
5.9
IN:
scriptSig A
OUT:
scriptPubB, 5.0
scriptPubA, 0.9
...
IN:
scriptSig A
OUT:
scriptPubC, 10.0
IN:
scriptSig ...
OUT:
scriptPubA, 9.2
...
CMU
Slide credit: Joe Bonneau

26. Module III: Learning Outcomes
Understanding of cryptography behind
Anonymous communication
Anonymous cash (zero-knowledge)

27. An Organizing Viewpoint
Privacy as a right to restrictions on personal information flow
Collection
Inference
Use
Dissemination

28. Privacy enhancing technology adoption
Some drivers
User/citizen trust
Ethics/culture
Internal team
Regulation
Public opinion

29. Thanks! Questions?