Authorship certificate

An authorship certificate is process evidence. Instead of judging a finished text by how it looks — which is what AI detectors do — it records the writing process itself: edit events collected as the document is written, chained into an append-only log, and digitally signed so the record cannot be altered after the fact. Anyone verifying the certificate can confirm the signatures are valid and see that the text accumulated through drafting and revision over time rather than arriving in a single paste.

The key contrast is verdicts versus evidence. A detector outputs a probability judgment about a finished text; a certificate documents how that text came to exist. The first can be wrong with no way to appeal. The second is checkable.

This matters most for non-native English writers, because they are the group detectors fail worst. A Stanford study (Liang et al., published in Patterns, 2023) found AI detectors falsely flagged 61% of TOEFL essays by non-native English speakers. A student facing a false flag usually has nothing to offer beyond "I wrote it." A signed process record changes the conversation from does this text look human to here is how it was written.

The honest limits: a certificate documents what happened inside a given editor, not outside it, and the writer has to opt in before writing starts — you cannot certify a document retroactively. It is evidence, not absolution.

Diglot implements this as its Authorship Certificate: an ed25519-signed edit-event chain generated while you write, which you can share with an instructor or editor for independent verification.