AI detectors by use case

An AI detector by use case is a filtered view of the same seventy-two cards, not a different set of claims. Each of the eight views in this hub names the two to four card fields that decide fit for one kind of reader, lists every card that states at least one of them, and quotes the value with its source and the date it was read. A business view reads the retention statement, the API and the team plan; a school district view reads the learning platform integration, the student data statement and the false positive evidence; an agency view reads input limits, bulk checking and price per word; a recruiter view reads the false positive evidence first; a code view reads which cards state code detection at all; an admissions view reads retention, false positive evidence and a 650-word essay in one check; a marketplaces view reads an API, a stated minimum under 300 characters and a bulk route; a researchers view reads a per-check limit of 5,000 words or more or a document upload. The fields are those of the detector index, the rules are on the methodology page, and the full board shows the cards side by side.

A view is deliberately not a recommendation. It does not say which detector a district should buy; it says which cards state an integration with Canvas or Moodle, which quote a false positive figure from a published study, and what each vendor's privacy policy says about student work, so that the reader can weigh those fields against a contract. Where a card states none of the deciding fields it is absent from the view, and the view says how many of the seventy-two qualified. On 8 November 2026 the school district view held six cards, the business and recruiter views seven each, the agency view seven and the code view two, because only Copyleaks and Originality.ai stated code detection. The six cards added on 9 November 2026 put Codequiry into the school district, agency and code views, Reality Defender into the business view beside Codequiry, and GLTR and the GPT-2 Output Detector into the recruiter view. Two views were added in the fifth wave on 9 November 2026: the marketplaces and review platforms view holds two cards, Phrasly and Sapling, the only ones stating all three of its deciding fields, and the researchers view holds nineteen cards with a hosted tool that take a full manuscript, read beside Gao 2023's result on medical abstracts.

Five views were added in the eighth wave on 9 November 2026, bringing the hub to thirteen: the law firms view holds 9 text cards that state a retention period or no-storage route and an enterprise deployment or report export; the grant reviewers view holds 18 cards that take a full proposal and state a retention policy, read after the NIH notice of 23 June 2023 and the NSF notice of 14 December 2023 on uploading proposals to outside AI tools; the newsrooms view holds 5 cards with an API or browser extension and a stated minimum under 500 characters; the localisation teams view holds the 14 cards stating 10 or more languages beside their non-English study records; and the government view holds 18 cards stating an on-premise deployment, a named data region or a security standard, with 0 stating FedRAMP.

A fourteenth view was added on 10 November 2026 in the tenth wave: the self-publishing view holds the 12 text cards that state a document upload or a per-check limit of 5,000 words or more and a retention policy, with the per-check limit, upload formats, retention clause and published price per 1,000 words for each, read beside the Kindle Direct Publishing Content Guidelines fetched live the same day, which require disclosure of AI-generated text and exempt AI-assisted text.

Evidence is read into each view from the studies index. The school district and recruiter views lean on Liang and colleagues, who reported that seven detectors flagged an average of 61.22% of 91 essays by non-native English writers as AI (DOI 10.1016/j.patter.2023.100779), because English learners and overseas applicants are the people most exposed to a wrong flag. The agency and business views lean on Perkins and colleagues, who found that simple edits to AI text lowered detection across seven tools (DOI 10.1186/s41239-024-00487-w), because edited drafts are the texts those readers check. Each view links the records it draws on and repeats the base of every figure.

The table below lists the views, eight on 8 November 2026 and thirteen since 9 November 2026, with their register IDs, how many cards each held on the date the view was read, and the fields that decide fit. The same cards appear in the four rankings, ordered rather than filtered; a reader who wants to compare several tools on one published field should start there, and a reader who wants to know whether a tool fits a job should start here.

Every page in this hub carries the same caveat as the cards: no detector score should be the only evidence in a misconduct, hiring or publishing decision. The vendors say the same in their own words, quoted on the cards, and the studies indexed here give the reasons.

A fifteenth view closed the roadmap on 10 November 2026 in the twelfth wave: the customer support QA view holds the two cards, Phrasly and Sapling, that state an API, a short-text statement under 500 characters, a bulk route and a retention policy together, with four more that state three of the four, read 10 November 2026.