Launched this week
GPTypo
A false positive shield for human writers
9 followers
A false positive shield for human writers
9 followers
AI detectors flag real human writing as AI more often than they should. GPTypo scans your text for the patterns they look for, shows which sentences trip them, and helps you fix each in your voice. On the web or the new Chrome extension.
Does it actually keep my voice intact after the rewrites, or does it tend to flatten everything into that generic AI-sounding tone I’m trying to avoid in the first place?
@nazlcanzay5vlt Definitely tries to keep your voice intact. It works sentence by sentence, so it only rewrites the ones a detector flags and leaves the rest of your writing alone, no smoothing the whole piece into a flat tone. It matches the tone of what you wrote and gives you a couple of editable options per sentence. Try pasting in something you wrote and see if it still sounds like you. If it doesn't, please let me know!
Does the extension check text as you type in Gmail and Google Docs, or only when you manually run a scan on selected content?
@edanur1076469 Only when you run it, not as you type. By design, it doesn't watch your keystrokes or run in the background. It's a tool for when you want a check, not something running always, and nothing leaves the page until you choose to scan.
Hey Product Hunt 👋
AI detectors flag real human writing as AI far too often. Roughly 1 in 20 pieces get caught, and it hits students, non-native English speakers, and anyone who writes in a clean, formal style the hardest. GPTypo is for writers who wrote every word themselves and want to catch a bad flag before someone else does.
Here's how it works. Scan your text and you'll see what a detector sees: a score, the exact sentences that trip it, and a reason behind each flag. Click any flagged sentence for a rewrite in your own voice. You can do this in the web app at gptypo.com, or with the Chrome extension right where you already write, no copying into a separate tab. The extension is the piece I'm launching today.
The free tier is open to everyone with no account, and if you want to see how well the detector actually holds up, our accuracy and false-positive numbers are at gptypo.com/transparency.
One more thing, given the timing. This week the company that owns Grammarly bought GPTZero, one of the big AI detectors, and is putting it inside a writing assistant that runs almost everywhere you type. So now the same company is helping you write and deciding whether your writing looks like AI. Even if everyone means well, you'd want a second opinion from someone who isn't on both sides. That's the job GPTypo does. It's independent, you come to it when you want a check, and we post our numbers on a shared public benchmark anyone can look up.
I'd love your feedback, and thanks for taking a look.
Forrest