Vega: Zero-Knowledge Proofs for Digital Identity in the AI Era
Key point
Vega has implemented ZKPs that prove only facts from government-issued credentials without revealing the original documents.
Details
Vega is a zero-knowledge proof system that proves only facts such as age, proof of personhood, and professional qualifications from government-issued credentials without exposing the original documents. Credentials never leave the device, and the system is designed to reduce privacy exposure even in authentication environments where AI agents are involved as intermediaries.
While governments and service providers have relied on ID upload methods, users have had to entrust sensitive documents to transmission, storage, and deletion processes. As policies such as the EU Digital Identity (EUDI) framework, age verification blueprints, and the UK Online Safety Act spread, the need for a verification method that satisfies both accuracy and privacy has grown.
Vega operates without trusted setup by directly combining Spartan, Nova, HyperNova, and NeutronNova. Written in Rust, it targets real-world formats such as mobile driver's licenses and the EU Digital Identity Wallet, and is set to be open-sourced soon.
Performance also emphasizes practicality.
- For a typical mobile driver's license (about 2KB), proof generation takes just 92ms
- Proof size is 108KB, with verification at 23ms
- The prover key is 464KB, small enough to fit comfortably on a smartphone
- For smaller credentials, generation takes 62ms, with an 83KB proof and 17ms verification
The key lies in fold-and-reuse proving and lookup-centric circuit design. Instead of unrolling the entire SHA-256, a step circuit that proves only a single compression step is created multiple times, folded together with NeutronNova, and then Spartan proves only the step and core circuits. Additionally, rather than building a full CBOR parser, byte-addressable lookups read only the necessary fields, hiding length and structure while maintaining efficiency.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.