privacy preserving e-petitions

1. privacy preserving e-petitions Claudia Diaz, HanneloreDekeyser, MarkulfKohlweiss, GirmaNigusse K.U.Leuven IDIS Workshop 29/05/2008 [Work done in the context of the ADAPID project]

2. ADAPID • Advanced applications for e-ID cards • Basic Research project funded by IWT • Focus on security and privacy • Belgian e-ID based on PKI with X509 certificates • Hard to build privacy friendly applications • E-Government • Hard to find applications where anonymity is arguably necessary

3. Petitions in the physical world • Formal request addressed to an authority and signed by numerous individuals • Typically citizens provide • Unique identifier (name, national ID number) • Signature • Verification: • Validating that the signatures correspond to the identifiers • Discarding multiple/invalid signatures

4. Electronic petitions • Benefits of going electronic: • Many resources are needed in order to physically collect the signatures • Manual signature verification is a costly and tedious process • Drawbacks • Easy to cheat (e.g., knowledge of other people’s name and ID number) • Countermeasures may “disenfranchise” petition signers (e.g., IP detection) • Good example of ICT enabling participatory e-democracy

5. The naive e-petition implementation • Have users sign the petitions with their e-ID • Select petition • Sign using the e-ID (2-factor authentication) • Validate signature and check that the petition has not yet been signed with that e-ID • Count (or discard) the signature • Privacy risks • Leak sensitive information on political beliefs, religious inclinations, etc. (this may prevent some people from signing) • Through unique identifiers, petition signatures can be linked to other data

6. e-petition requirements • Basic requirements • Authentication: citizen is who claims to be (i.e., no impersonation) • Authorization: citizen is entitled to sign (e.g., age > 18) • Integrity, Confidentiality • Multiple signing prevention • Verifiability: all valid signatures are counted • Privacy requirements • Signer anonymity: citizen unlinkable to petition (i.e., not possible to identify who are the signers)

7. Anonymouscredentialprotocols • Active area of research in cryptography • Theyrely on cryptographicprotocols and Zero-Knowlegeproofs to reduce to the bare minimum the amount of information disclosed • Flexible protocols, many options possible • Example: • CI issues a credential to U that encodes U’sage • U canprove to V thathisageisabove/below a threshold • V can check thatthisiscertified by CI • V does not learnU’s exact age

8. PKI vsanonymous credentials PKI Anonymous credentials • Signed by a trusted issuer • Certification of attributes • Authentication (secret key) • Double-signing detection • No data minimization • Users are identifiable • Users can be tracked (Signature linkable to other contexts where e-ID is used) • Signed by a trusted issuer • Certification of attributes • Authentication (secret key) • Double-signing detection • Data minimization • Users are anonymous • Users are unlinkable in different contexts

9. Architecture

10. Properties • Only citizens entitled to sign can do so • Possession of e-ID + knowledge of PIN • Attribute verification (e.g., age, locality) • One credential per citizen • Citizens can sign only once (multiple signing is detectable so that repeated signatures can be deleted) • Collusion of credential issuer and e-Petition server does not reveal the identity of a signer • Verifiability through publishing the protocol transcripts

11. Summary and conclusions • Summary of the paper • Motivation for privacy preserving e-petitions • Requirement study • Introduction to anonymous credentials • Architectural design combining existing technologies • Legal issues • To be added: details of the protocols and implementation • Proof-of-concept • We can satisfy seemingly contradictory requirements • Security properties can be achieved without identifiability • National ID can be user to bootstrap privacy friendlier IDM, while preventing Sybil attacks

12. Thankyou!