Auberon LLC

Scan selected text for AI detection patterns and get sentence-level rewrites.

Scan any text on the web for the patterns AI detectors flag, then fix them sentence by sentence, without leaving the page. GPTypo is independently benchmarked on RAID (ACL 2024, raid-bench.xyz) with 99.24% accuracy and a 0.5% false-positive rate on real human writing. 🪶 Quick start 1. Highlight text on any page. 2. Click the floating "Scan with GPTypo" button, or right-click → "Scan with GPTypo." 3. See a 0–100 detection score with a sentence-level breakdown in the side panel. 👉 What you can do Select any text on any webpage and GPTypo will scan it. You'll see exactly which patterns triggered each flag: uniform sentence lengths, AI-tell phrases, overused transitions, overly formal phrasing, and more. Click a flagged sentence for an AI-powered rewrite tuned to your tone, then replace it inline on editable surfaces or copy it out. Pro users can generate a shareable proof report: a public URL that demonstrates the text was scanned and scored. 🎯 How it works The free tier runs an instant scanner on text up to 500 words, useful for quick drafts. Pro and Team plans unlock verified scoring from our fine-tuned neural detection model, the same one ranked publicly on RAID. Methodology and accuracy data are open at gptypo.com/transparency. GPTypo is built for human writers worried about false positives, not for bypassing detectors. Text you scan is sent to our API for scoring and is never sold. Full privacy details at gptypo.com/privacy. Questions or feedback? support@gptypo.com