I got tired of AI text getting flagged by GPTZero and Originality, so I reverse-engineered what detectors actually scan for. Then I built a free scorer that shows your text's "AI-risk" % instantly.
What detectors look for
- Long, uniform sentences. A steady 20-word cadence is a dead giveaway. Human writing varies — short punch, long explanation.
- AI filler vocabulary. Words like delve, leverage, robust, seamless, navigate the complexities spike the score.
- Zero rhythm variance. No fragments, no dashes, no commas doing natural pauses.
The free fix
I turned the whole thing into a free tool + a prompt pack:
- Free AI-Risk Score — paste text, get an instant % and the exact sentences to fix: https://cc6a24a943e446d38bf5cac7200559bc.app.codebuddy.work/tools/ai-risk-scorer.html
- Prompt packs that humanize drafts — the exact prompts I use: https://felovery.gumroad.com (launch code FIRST50 = 50% off everything, no expiry)
Try it
Paste one of your AI drafts into the scorer. If it's over 60%, run it through the prompt pack and re-score. The drop is usually 30+ points.
Hope it helps — happy to score a sample in the comments.
Top comments (0)