Cryptography · Long-horizon evidence

Post-quantum audit trails, and why retention changes the calculation

Most cryptographic decisions are made against today's threat model. Audit evidence is different: it has to survive a retention period measured in years, against an adversary who gets to choose when to attack.

Encryption and signatures are usually evaluated against current capability: is this algorithm sound today? For data in transit, that is the right question, because the exposure ends when the session does.

Audit evidence is the opposite case. The record has to hold up for as long as it is retained, and the party who might want to forge or repudiate it gets to choose the moment they try.

The retention mismatch

  • EU AI Act Article 18 requires technical documentation to be kept for ten years.
  • Financial recordkeeping obligations commonly run five to seven years, longer under litigation hold.
  • Disputes about automated decisions can surface years after the decision.

A signature made with classical elliptic-curve cryptography is sound now. The relevant question for a ten-year record is whether it will still be sound when someone finally challenges it.

Harvest now, decrypt later

The phrase describes an adversary who collects protected material today and waits for capability to catch up. For confidentiality the risk is disclosure. For signatures the risk is different and arguably worse: an adversary who can forge a signature after the fact can manufacture records that appear authentic, or repudiate genuine ones.

For an evidence system, that is the failure that matters. Evidence whose integrity guarantee expires quietly, mid-retention, is worse than no evidence, because nobody is watching for the moment it stops being true.

What a post-quantum audit trail looks like

Rubric signs attestations with ML-DSA-65, the NIST-standardized module-lattice digital signature algorithm, and anchors the signed records to a public ledger for independent timestamping. Two properties follow:

  • The integrity of the record does not rest on assumptions about factoring or discrete logarithms.
  • The time of the record is witnessed externally, so even a future cryptographic break does not let someone quietly insert a backdated record into the sequence.

Check it yourself

Every Rubric attestation resolves publicly, with no account and no API key, and every anchor resolves to a public ledger message you can read without our cooperation.

HCS topic 0.0.10416909 · ML-DSA-65 signatures

Verify an attestation · Read the ledger ↗

Layering with existing standards

Post-quantum signing does not require abandoning the standards ecosystem. Rubric's evidence documents carry C2PA Content Credentials, which use conventional signatures as the ecosystem specifies, wrapped around a post-quantum attestation layer underneath. Standards-compliant envelope, quantum-durable core.

Verifiable, not asserted

Rubric Protocol is a conformant C2PA Generator Product (Content Credentials 2.4, Assurance Level 1).

Record 01a002b7-3663-7b3b-a60e-db3b99ee2d94 · Echelon Intelligence Group LLC

Primary-source records

Migration is not an event

Most organisations will migrate to post-quantum cryptography gradually, driven by vendor timelines they do not control. Evidence is a good place to start early, precisely because the retention horizon means records created today are the ones most exposed to whatever arrives later.

Related: Tamper-evident AI logs · How to prove an AI decision happened · Key custody