Crescent protects the privacy of digital identities | Keryc
Microsoft Research introduced Crescent, a cryptographic library designed to prevent your digital identities from being tracked when you use them in apps or on your phone. Have you ever felt like when you use a mobile credential someone is following you from service to service? Crescent aims to cut that thread by preventing the linkability between different uses of the same credential. (microsoft.com)
What Crescent is and why it matters
Crescent is a library that adds unlinkability to already-popular identity formats, like JSON Web Token and mobile driver's license credentials. The core idea is simple: beyond what you choose to reveal, nothing inside the credential itself should let different presentations be linked together. That lowers the risk of detailed profiles being built about your behavior. ()
And the best part? It doesn't force credential issuers to rebuild their infrastructure. Crescent can be integrated without asking departments or organizations that issue documents to change their systems. (microsoft.com)
In short: protect privacy without requiring a massive rewrite of the identity ecosystem.
How it works in short
Crescent uses zero-knowledge proofs, specifically a form called Groth16, to prove you have a certain attribute from a credential without revealing the whole credential. This kind of proof allows verifiers to check information compactly and quickly. (microsoft.com)
To make it practical on mobile, Crescent splits the process into two stages:
prepare: a phase that runs once and generates cryptographic values stored on the device.
show: when you need to present your credential, this phase uses the prepared values and randomizes them to avoid any link with previous presentations. The result is a short proof that reveals only what’s strictly necessary. (microsoft.com)
Also, the mathematical rules that define those proofs are translated into a system called R1CS, which describes exactly what must be checked without leaking extra data.
A real example: employment and age verification
In Microsoft's demo they built a sample app with two practical scenarios. First, employment verification: you prove you work at a company (for example Contoso) to access benefits without the verifier learning your full identity or your employer learning about the transaction. Second, age verification: you show you're over 18 without revealing your date of birth or identity. In both cases, the presentations remain unlinkable. (microsoft.com)
Sound useful for services like healthcare, social networks, or events? Exactly. It prevents a single attribute from acting like glue that tracks your movements.
What changes for companies and users
For users: more control over what you share and less chance of invisible profiles built from hidden identifiers.
For companies and governments: the ability to add privacy protection to existing credentials without replacing them immediately. There will be engineering work to integrate clients and verification, but the barrier to entry is smaller than having to standardize new signatures across every system. (microsoft.com)
Where to find Crescent and how to try it
Microsoft published the project on GitHub and accompanied the announcement with a sample app and conference presentations. If you're a developer or identity lead, the repository includes example code in Rust, a browser wallet, and documentation to generate parameters and proofs. (microsoft.com)
Final reflection
Digital identity privacy moves from promise to concrete tools. Crescent shows that with modern techniques like Groth16 and a two-stage design, it's possible to give users more control without freezing the ecosystem we already use. Does this mean the end of credential-based tracking? Not overnight, but it's a practical, realistic step toward systems where you decide what gets shared and what doesn't. (microsoft.com)
Stay up to date!
Get AI news, tool launches, and innovative products straight to your inbox. Everything clear and useful.
Crescent protects the privacy of digital identities
