Jonas Nick of Blockstream Research and collaborator remix7531 have released libshrincs, a proof-of-concept C library that implements a key component of a quantum-resistant signature scheme designed for Bitcoin. The library, announced on August 11 on Delving Bitcoin, represents one of the most rigorous attempts yet to formally verify cryptographic code meant to shield Bitcoin from the eventual arrival of powerful quantum computers.
The project is notable not just for what it protects against, but for how it was built. The security and correctness proofs were largely generated with the help of large language models, including ChatGPT and Fable, under human oversight.
What libshrincs actually does
Libshrincs implements WOTS+C, the one-time signature component of a broader scheme called SHRINCS. SHRINCS, first proposed in December 2025, is a signature scheme that blends stateful and stateless hash-based signatures. The goal is to keep signature sizes compact enough to be practical for Bitcoin transactions while still providing quantum resistance.
In stateful mode, signatures clock in at roughly 324 bytes. The scheme also includes a stateless fallback mechanism, essentially a safety net in case the stateful component’s tracking gets mismanaged.
The library ships with machine-checked proofs of functional correctness and a security proof verified using SSProve in Rocq. The proofs are substantial: approximately 6,200 lines of VST (Verified Software Toolchain) code and around 13,400 lines of Rocq code.
The AI angle, and why it matters
Large language models were used to generate substantial portions of the verification code, with human developers guiding and reviewing the output. Formal verification in cryptography has historically been expensive and slow. If AI-assisted methods can reliably produce this kind of output, it could dramatically lower the barrier to building verified cryptographic software.
Limitations and the road ahead
The researchers are transparent about what libshrincs does not yet accomplish. The current security theorem does not provide a complete post-quantum bound in the quantum random oracle model (QROM). It also uses a weaker security notion relative to how SHRINCS signatures would actually be deployed in practice. Numerical hash assumptions that would strengthen the security argument have not yet been incorporated.
There are no claims of production deployment or integration into Bitcoin Core. The work sits within Blockstream Research’s broader initiative to develop quantum-resistant signatures for both Bitcoin and Blockstream’s Liquid sidechain network. Related efforts include C++ implementations and Simplicity verifier libraries.