A tool sits in exactly one category - the job it is sold for. Tools that do two things, like the vendors who sell both a humanizer and a detector, are filed under the one they lead with, and the listing says what else they do.
Tools whose main job is taking AI-generated text and rewriting it so a human reader — and, they claim, a detector — takes it for human writing. This is the category the whole field is named after, and the one where vendor claims and measured results diverge the most.
Products positioned first and foremost around beating a named detector rather than around writing quality. The distinction matters: a tool marketed on evasion is optimised against the detectors that existed when it shipped, which is why a dated test result is worth more here than anywhere else.
Classifiers that estimate whether a passage was written by a language model. Every one of them produces false positives on human writing, and none publishes an independently audited error rate, so we record the vendor's own accuracy claim and mark it as exactly that.
Rewriting tools that predate the detector arms race: synonym swapping, sentence restructuring, tone changes. Many are used as humanizers, but they are built for clarity and length rather than evasion, and they usually say so.
Similarity engines that compare a document against the web, journals and student archives. Plagiarism detection and AI detection answer different questions, and conflating them is the most common mistake in this field.
Integrity platforms bought by a department rather than a person: LMS integration, class-wide reporting, proctoring. Pricing is usually sales-only, which is itself a fact worth listing.
Products where the developer API is a first-class offering: documented endpoints, per-word or per-credit pricing, rate limits you can plan around. Relevant if you are putting a humanizer or detector inside your own product.
Tools marketed specifically at students and academic writing. We list what each one does and what it costs. Whether using one is permitted is set by your institution's rules, not by a directory — see our note on academic integrity.
Humanizers and rewriters sold to content teams producing at volume. The buying question here is rarely detection alone: it is whether the output survives an editor, and what the per-article cost works out to.
Tools that explicitly support languages other than English, or are built for one specific non-English market. Detector accuracy drops sharply outside English, and so does the quality of most rewriting — both are things we test for separately.
Broad AI writing platforms that ship an explicit humanize, de-AI or 'sound human' feature alongside everything else. Listed here only when the feature is the vendor's own, not a third-party integration.