Digitally signed documents in the quantum era: preserve, don't re-sign

Will banks and other regulated industries have to re-sign every digitally signed document before quantum computers can break the classical algorithms that signed them? The instinct is to say yes and scope a very large re-signing exercise.

The answer is: not necessarily.

Documents signed today with RSA or ECDSA rely on algorithms that would become forgeable once a cryptographically relevant quantum computer exists. For records that must remain legally provable for years, that is a real integrity problem. A signed record retained for ten or twenty years may outlive the security of the algorithm that signed it.

NIST's current transition timeline puts that crossing in view: quantum-vulnerable signature algorithms at the 112-bit security level are proposed for deprecation after 2030, with RSA and elliptic-curve signatures disallowed after 2035.

Re-signing the whole archive is the obvious response, but it should not be the default. It is expensive, difficult to scale across millions of records, and introduces a new signing event rather than preserving the provenance of the original one. A fresh signature attests to the moment of re-signing, not to who signed the document originally or when.

Instead, preserve the existing signatures. You validate the signatures while their algorithms are still trustworthy, preserve the documents and their validation data, and protect that evidence with renewable archive timestamps. Ultimately using quantum-safe algorithms.

That does not stop the original algorithm from becoming forgeable. It preserves evidence that the document and its signature existed, and were considered valid, before that happened.

Timing is the load-bearing part. The preservation evidence has to be created while the classical signature can still be trusted. The method can also work in bulk: one timestamp over a hash tree can protect a large batch, and the evidence is renewed over time as algorithms weaken.

The mechanism has been standardised by the IETF since 2007. RFC 4998 defines archive timestamps and their renewal to preserve evidence as signing, hashing and timestamping algorithms weaken.

In the EU, qualified preservation is a regulated trust service under eIDAS, while ETSI TS 119 511 defines policy and security requirements for preservation-service providers.

The Cloud Security Alliance recommends the same general approach for long-lived signed documents: extending eligible signatures to archival formats such as PAdES-LTA and renewing their timestamps.

Real archives rarely contain one clean signature format. They contain PAdES, XAdES, CAdES and CMS-based signatures, often in legacy profiles with incomplete validation data or unavailable OCSP and CRL sources.

Preservation means identifying which signatures can still be validated, reconstructing the evidence that supports them, applying appropriate preservation mechanisms, aggregating records efficiently and renewing the resulting evidence before its cryptography becomes weak.

At CENSUS, we design and build this. We find where signatures and their dependencies live, integrate the cryptography, and engineer the preservation and verification pipelines that keep evidence provable across the post-quantum transition.

• RFC 4998, Evidence Record Syntax. IETF, 2007.
• ETSI TS 119 511, long-term preservation of digital signatures.
• Cloud Security Alliance, Post-Quantum Preparedness. 2024.
• BIS Project Leap Phase 2. BIS, 2025.