What Authentext can't do
Every tool in this category has limits. Most don't publish them. Here are ours — plainly, so you can calibrate your trust before you extend it.
1. It cannot prove authorship.
No software can. Authentext produces evidence — measurable departures from a student's own baseline — never proof. Any consequential decision requires human judgment, context, and conversation. A tier is where a conversation starts, not where one ends.
2. Lightly edited AI text is hard to catch.
A student who writes a real draft and uses AI to polish a sentence or two sits close to legitimate revision — for us and for every tool. Our per-student baseline notices more than generic detectors do, but small blends are the hardest case in the field, and we won't pretend otherwise.
3. Short texts carry less signal.
Stylometry is statistics; statistics need material. Very short submissions (a paragraph, an exit ticket) produce lower-confidence readings, and Authentext says so on-screen rather than bluffing.
4. New students start with thin baselines.
Until enough verified writing accumulates (roughly ten samples), confidence is reduced and the system deliberately errs conservative. Importing past work closes this gap faster; it doesn't skip it.
5. Real voices change — and that can flag.
A new genre, a growth spurt, a register experiment can surface as an anomaly precisely because it's new. That's by design: the tool flags departure, and the teacher supplies the context that resolves it. Expect occasional benign flags; they take about a minute to clear.
6. A sufficiently determined student can evade any tool — including this one.
Sustained, disciplined imitation of one's own style is the attack no system in any era has fully solved. What Authentext does is raise the cost of cheating above the cost of writing, watch for the signatures imitation leaves, and create evidence when it's attempted. Deterrence for the many, difficulty for the few, certainty for no one — that is the honest ceiling of this entire category.
7. It is not disciplinary evidence on its own.
Authentext output should never be the sole basis for an academic-integrity finding. It informs a professional's judgment; it doesn't replace it.
If anything else on this site ever appears to promise more than this page, this page wins.