Changelog
What we’ve changed and when.
Public record of meaningful updates to the platform — new detectors, blog additions, methodology changes, and policy revisions. Subscribe via your feed reader of choice through the site’s sitemap.
- Tools
Launched AI Humanizer and Humanization Detector
Two new tools added to the platform. The AI Humanizer (/ai-humanizer) rewrites LLM output into more natural prose with Light/Medium/Aggressive levels — contractions, transition swaps, sentence restructuring, vocabulary downgrades, and pacing inserts. The Humanization Detector (/humanization-detector) catches text that has been processed by a humanizer using a signal model targeting register collisions, characteristic fillers, and engineered sentence-length variance. Both run entirely in the browser. Redirects added: /humanize-ai, /humanizer, /humanize-text, /undetectable-ai all 301 to /ai-humanizer.
- Content
Keyword-driven SEO expansion (text + image + brand alternatives)
Tightened the text-detector, image-detector, and ai-essay-detector metaTitles around real high-volume search phrasings. Added /ai-art-detector (4.4K monthly volume) and /ai-detector-for-teachers (3.6K). Published four brand-alternative comparison pages — GPTZero, Turnitin AI Checker, Originality.ai, Winston AI. Published four new blog posts including does-turnitin-detect-ai (6.6K monthly direct query) and an honest editorial cluster on AI humanizers and undetectability claims.
- Content
Replaced /platforms with /social-media-ai-detection
The previous /platforms hub had no clear search-keyword target. Rewrote and renamed to /social-media-ai-detection with a sharper H1 and metadata aimed at real social-media AI detection queries. /platforms now 301-redirects to the new URL.
- Content
Audience pages, comparisons, glossary, and 2026 statistics
Added five audience-vertical pages (journalists, educators, marketplaces, HR, parents), four comparison pages, a 16-term glossary, and a curated 2026 detection-statistics reference page. Added blog category index pages.
- Policy
About, privacy, methodology, contact, and terms pages
Published canonical trust pages. The privacy page documents the browser-side processing model. The methodology page documents per-content-type signal models and explicit detection limits.
- Tools
Per-content-type uploaders
Replaced the unified upload widget with six specialized interfaces — image drop-zone with thumbnail, video URL-first with platform examples, text large-textarea with word counter, audio file plus inline playback, music file with track context, screenshot kind selector with preview.
- Content
Blog launch
Published the first 13 guides across four categories — detection guides, platform safety, AI explainers, scam awareness — each linked to a primary tool and a set of related guides.
- Methodology
Softened classification labels
Replaced 'Real / AI-generated / Manipulated' with 'Likely authentic / Likely synthetic / Likely manipulated / Inconclusive' to reflect the inherently probabilistic nature of detection. Added an explicit demo-mode notice on every result.
- Content
Per-page unique content fields
Replaced the templated tool-page sections with hand-written intro, how-it-works, when-to-use, limitations, and FAQ content per detection page.
- Policy
Title cannibalization and FAQ duplication cleanup
Consolidated overlapping pages with 301 redirects, gave each page a unique meta title, and replaced the templated FAQ block with hand-written per-page questions.
For the full editorial standard see about · For privacy changes see privacy · For what each detector does, see methodology.