About HumanizerBench
HumanizerBench is an independent monthly benchmark of every major AI humanizer on the market. Each cycle, we run the same set of prompts through every tested humanizer, submit each humanized output to five commercial AI detectors, score the results with a published deterministic formula, and ship the raw data (inputs, outputs, detector scores, scoring script) alongside the leaderboard.
The market for AI humanizers is noisy. Every vendor claims the highest bypass rate against the most detectors with the best meaning preservation. We exist to give that market one credible, auditable reference: the same inputs, the same detectors, the same scoring, every month. If we say a humanizer ranks third, you can download the cycle and reproduce the rank yourself. The longer version of why we built this goes into more depth.
We're operated by WriteHuman, which sells one of the humanizers on this leaderboard. We disclose that on every page and we hold WriteHuman to exactly the same methodology, scoring, and penalty rules as every other tested humanizer. Total transparency is the only credibility play that works when the operator is also a competitor.
Operator
HumanizerBench is operated by WriteHuman (https://writehuman.ai). WriteHuman builds an AI humanizer for writers, students, and marketing teams. Editorial control of the benchmark (what we test, how we score, what we publish) is independent of WriteHuman's commercial product team.
Anti-bias commitments
- WriteHuman is tested using the same methodology and scoring as every other humanizer.
- The scoring script is public and deterministic.
- We do not adjust rankings outside the published methodology.
- We do not accept payment for placement.
- We do not exclude tools because they outperform WriteHuman, and we do not run private tests off the record.
The full version of these commitments, and how to dispute a result, is on the fairness & corrections page.
Contact
- Press: contact@humanizerbench.com
- Partnerships: contact@humanizerbench.com
- Disputes & corrections: contact@humanizerbench.com (see fairness for the full process)
Open source
All inputs, outputs, detector scores, and the scoring script live in our public repository: https://github.com/HumanizerBench/humanizerbench.