Glossary
Non-Repudiation
Evidence that holds up later: who sent what, who received it, and when, recorded in a way the parties cannot credibly walk back.
General definition
Non-Repudiation is one of the classic security services alongside confidentiality, integrity, authentication and availability. It means a party to a transaction cannot successfully deny their involvement afterwards. ISO/IEC 13888 splits it into non-repudiation of origin (proof the sender created and sent the message), non-repudiation of delivery and receipt (proof the message reached the recipient and was accepted), and related forms for submission and transport. The evidence has to convince a third party, typically a court, an arbitrator or a regulator, which rules out anything the operator or either party could have fabricated alone.
The mechanisms stack:
- Digital signatures created with a private key only the signer holds (RSA, ECDSA, Ed25519) prove origin and bind the signer to the exact content; their legal weight comes from frameworks such as eIDAS in the EU and the ESIGN Act in the US
- Trusted timestamps (RFC 3161) from a time-stamping authority prove the content existed at a point in time
- Authentication and audit trails tie every action to an identified account, device and session, which in most business systems is the practical form of evidence
- Delivery and read evidence: server-recorded acknowledgements, delivery receipts and read receipts show that a message reached a device and was displayed
- Immutable storage: hash chains, write-once media and WORM storage make it detectable if a record was altered after the fact, which is what gives the rest of the chain credibility
In messaging, non-repudiation serves compliance more than cryptography. Financial regulators require firms to keep complete, unaltered records of business communications (SEC Rule 17a-4, FINRA rules), healthcare rules require an accurate record of who accessed and sent protected information, and disputes about instructions, consent or notice turn on being able to prove the message was sent, delivered and seen. A plain audit log that nobody can edit, with synchronised clocks and strong login, achieves most of this; per-message signatures are added when the counterparty is outside the organisation or the stakes justify it. Note the tension with privacy: end-to-end encrypted messengers often use deniable authentication on purpose, so that a transcript cannot be used to prove who said what, which is the opposite design goal.
In the Ethora ecosystem
Ethora provides the server-side form of non-repudiation that regulated teams rely on. In the default compliance-archive model every message is stored in a server-side archive with its sender, room and timestamp, and the compliance audit trail records administrative and access events with immutable log export to S3, where an object-locked bucket in your own account makes later alteration detectable. Delivery and read receipts give evidence that a message reached a device and was displayed, and they are recorded by the server rather than reconstructed from client memory.
Identity is anchored through authenticated sessions: short-lived JWT chat tokens, social sign-on via OAuth and OIDC providers, and API-provisioned users tied to accounts in your own backend. Ethora does not add per-message digital signatures on the client; the evidence is the operator-held record, which is the model supervisory rules expect. Optional end-to-end encrypted rooms, switched on per app, reverse this trade: the server holds ciphertext it cannot produce as readable evidence, and only metadata such as sender, room and timing remains in the record.