Editorrah makes a signed record of how a piece of work was made: who wrote it, how much AI helped, whether its facts check out, and who signed it off. A university, a court, a journal or an employer can open that record and rely on it, instead of guessing from the finished text.
Live Two records are available today: one for universities and one for researchers. Five more are Next or on the Roadmap. See who it is for
30 minutes. We show a real record and say plainly what you can use today.
Editorrah Authorship record
Sample
Process timeline
4 sessions. 6 h 12 min active. 212 revisions.
Verification
| Authorship match | Consistent with 6 verified prior works | Verified |
|---|---|---|
| AI assistance | 2 segments detected, both declared | Declared |
| Fact verification | 14 of 15 references verified at source | 1 unverifiable |
| Human sign-off | J. Doe, Course instructor | Signed |
Sample record. Illustrative data, not a real submission.
The full record, shown as a sample. See what you can use today.
01Who it is for
Live: in use today.Next: being built.Roadmap: planned.
02What do I get?
01
Authorship match: consistent with 6 verified prior works
We read how it was written: the drafts, the edits, the time spent. Then we compare the writing with the author's own earlier verified work.
02
AI assistance: 2 segments detected, both declared
The author declares where AI helped. We check that against how the work was actually made, and show both side by side.
03
Fact verification: 14 of 15 references verified at source
Each claim, quote, citation and link is checked against its source. Anything we cannot check is marked unverifiable.
04
Human sign-off: J. Doe, Course instructor
A named person reviews the record and signs it. The decision stays with that person. The record is their evidence.
03How does it work?
Today the work is written in the Editorrah editor, which is how we see the way it was made. Three views show it.
01
Replay the work being written, from the first word to the last. Every paste is marked on the timeline, so you can jump straight to it.
Early feedback changed how students revised. Similar effects appear across disciplines.Pasted
02
Stylometry compares the writing with the author's own writing profile, so work that does not read like them stands out.
Whole piece: Needs review
03
For teachers, text pasted from outside is highlighted in red, outside text retyped by hand in amber, and sections flagged for AI-written content are boxed. On a co-authored paper, each passage is tinted by the author who wrote it.
Feedback works best when students can act on it quickly. Formative assessment improves outcomes in most settings.External Early drafts show the change. Peer review adds a second perspective.Retyped
Section 3Flagged
Large language models can now produce fluent, well-structured arguments on demand, which changes how written work is assessed.
Live In the Editorrah editor today, for universities and researchers. An API for other software is Next. Plugins for Canvas, Moodle and court e-filing systems are on the Roadmap.
More on the platformDevelopers: the same checks are coming as an API. Request access.
04What should I know?
05Where is the detail?
Why now
Since September 2025, five governments, from China to California, have put rules in force that AI-made content must be labelled. None of them says how a person proves what they made. The rules, the figures and the argument are on one page.
Why AI governanceChecked against the published instruments on 4 October 2026. Full register on the Why AI governance page.
Research
Our standards are set with academic integrity researchers. Our methods, and the limits of what a record can claim, are on the research page.
Read the researchSecurity
Data is encrypted in transit and at rest. Records are signed, so any later change shows.
Security overview