Microsoft Research has published Vega, a cryptographic system that uses zero-knowledge proofs to let people prove specific facts from a digital credential without revealing the credential or the underlying identity data. The work is aimed at digital identity use cases including the European Union’s digital identity wallet, AI agents acting for humans, and on-chain verification.
With Vega, a holder of a credential such as a mobile driver’s license can prove a fact drawn from it, for example that they are over a given age or hold a particular professional status, while the credential itself stays on the user’s device. A verifier receives only the proof of the specific fact, not the document or the other data it contains.
The research is led by Srinath Setty, Senior Principal Researcher at Microsoft Research, with co-author Darya Kaviani, and is set to be presented at the IEEE Symposium on Security and Privacy. Vega is designed to run on ordinary mobile hardware, generating a zero-knowledge proof in under 100 milliseconds with no trusted setup, a configuration step that has been a weakness in some earlier proof systems. For a typical 2-kilobyte mobile driver’s license, the team reports a proving time of 92 milliseconds, a proof size of 108 kilobytes, and a verification time of 23 milliseconds.
The system introduces several techniques to reach that performance, including a fold-and-reuse approach that lets repeated credential presentations skip expensive computation after the first proof, a lookup-centric circuit design that treats a credential as a byte-addressable table rather than parsing complex document formats, and device binding that signs session values with hardware-held keys so a proof can be tied to the user’s device. Vega builds on a line of proof systems developed in the research community, including Spartan, Nova, HyperNova, and NeutronNova.
Microsoft Research connects Vega to three settings. It targets frameworks such as the EU Digital Identity Wallet and its age-verification blueprint, where selective disclosure is a core requirement. It also points to AI agents, which could present identity proofs on behalf of a person without holding that person’s credentials, and to on-chain systems, where a user could prove a fact from a government credential to a decentralized service without an intermediary seeing the original document.
Vega is research rather than a product. It is implemented in the Rust programming language and is to be open-sourced, and its underlying proof system is already published.
Sources: Microsoft Research
–
By the ID Tech Editorial Team
