EditLens — Quantifying the Extent of AI Editing in Text

EditLens scores text on a continuous scale according to how much AI intervention it contains, rather than a simple human-vs-AI binary. This demo runs the RoBERTa-large model from the EditLens ICLR 2026 paper (code · model).

Paste a passage (≥ 50 words works best) and get an AI-editing score in [0, 1].

Model and dataset are licensed CC BY-NC-SA 4.0. EditLens is a research system and can be wrong — do not use it to make consequential judgments about real people.