Course contentsShow
← Security
Lesson 2928 of 3,05254. Privacy EngineeringFree lesson

Private Information Retrieval

Querying databases without revealing which records were accessed to the database operator.

Private Information Retrieval

What you'll learn: How to query a database without revealing which specific record you're accessing—even to the server holding that database.

The Privacy Problem in Database Queries

Imagine you're researching a sensitive medical condition and want to look up information in a health database. Normally, when you query a database, the server knows exactly which record you requested. This metadata leaks your interests, potentially compromising your privacy even if the query itself is encrypted in transit.

Private Information Retrieval (PIR) solves this by letting you retrieve a database record without the server learning which one you accessed.

How PIR Works (Conceptual View)

Think of a library where you want to borrow a specific book without the librarian knowing which one. In a naive PIR approach, you could ask the librarian to give you every book, then pick the one you want at home—perfectly private, but absurdly inefficient.

Real PIR schemes are smarter. They use cryptographic techniques (often building on homomorphic encryption or secure multi-party computation) to:

  1. Encode your query so it mathematically touches all records but only "activates" the one you want
  2. Process the query on the encrypted data without the server learning your selection
  3. Return a response that extracts only your desired record

Some PIR systems use multiple non-colluding servers: you split your query across them so that no single server learns your choice, but combining their responses gives you the answer.

The Tradeoff

PIR provides strong privacy guarantees but comes with computational and communication overhead. The server must often process more data than a direct query would require, making PIR most practical for high-privacy scenarios like anonymous credential checking or censorship-resistant content retrieval.

Key Takeaway: Private Information Retrieval enables querying databases without revealing access patterns to the server, protecting user privacy through cryptographic protocols at the cost of increased computation.