Detecting-AI.com AI detector

The free detector box reads "0 words · 0 / 15,000 characters" and accepts pasted text or TXT, PDF and DOCX uploads, with the FAQ noting that "PDF and DOCX text is read privately in your browser only after you choose a file" and that scanned PDFs are not supported (detecting-ai.com, read 10 November 2026). Every paid plan "includes AI detection for documents up to 1,000,000 characters per scan" with unlimited checks. The API and the MCP server take up to 160,000 characters a request, with usage counted in whitespace-separated input words and an Idempotency-Key that returns a stored result on an exact replay within 24 hours (detecting-ai.com/api-docs, read 10 November 2026).

Prices below are as displayed in USD on 10 November 2026 on detecting-ai.com/pricing, whose default Yearly toggle shows a per-month equivalent beside the annual total; the Monthly toggle's figures are in the page data. Payments run through Paddle, and the page states that local taxes may be added at checkout.

The API pricing page states that "An API key is free to create. Requests need an active subscription and remaining words" and that the page "lists the current monthly starting plans only. Do not assume an annual discount or custom rate" (detecting-ai.com/api-pricing, read 10 November 2026). The index's API price field converts $15 for 100,000 words to $0.15 per 1,000 words. Beyond the API, the vendor publishes a remote MCP server, a Zapier app with a Detect AI Generated Content action and a Moodle integration page describing an API-based connector with a server-side key, scoped with each institution before implementation (detecting-ai.com/integrations/moodle, read 10 November 2026).

Retention has no fixed period for detector submissions. The privacy policy's retention section reads: "Anonymous submitted content is not added to account history. It is processed to return the result and may remain briefly in necessary security or operational records maintained by us or the processor", and for signed-in users: "Signed-in scan history is kept while needed to provide account history ... Do not assume it disappears after a fixed period unless the relevant feature states one" (detecting-ai.com/privacy-policy, effective October 10, 2026, read 10 November 2026). The same policy states that "Your text is not sent while you type" and that the Humanizer Autopilot keeps its input and result for seven days. The retention rank field stays empty because "briefly" is neither a stated zero nor a period.

The home page's accuracy block reads "99% (accuracy), with under 1% false positives", with two tiles of "99% Detection accuracy" and "<1% False positive rate", and the API page repeats "99% detection accuracy on diverse datasets", "<1% false positive rate on human writing" and "1B+ human and AI texts tested" (detecting-ai.com and detecting-ai.com/ai-detector-api, read 10 November 2026). The API page describes the method as "A feature-based classifier and a fine-tuned transformer ... calibrated into one score, keeping wrong flags on human writing under 1% in Detecting AI's published benchmarks". No page read names the dataset, the count of human texts or the date behind the figures, so the claim is quoted and the false positive bound is not ranked.

The vendor's caveats are unusually direct. The method block states: "A model score cannot establish authorship, intent or misconduct. Treat it as one input alongside drafts, sources and conversation", and the FAQ: "AI detection is a probabilistic review signal, not proof of authorship or misconduct" (detecting-ai.com, read 10 November 2026). On false flags: "It happens, and it does not mean you did anything wrong. Formulaic structure, stock phrasing and very uniform sentence rhythm can resemble model output, especially in short samples." On edited text, the page says mixed drafts show "Different signals highlighted in context" and that "Short, translated, heavily edited or formulaic writing can be harder to classify", while the same vendor sells a Humanizer Autopilot that "rewrites and checks each section with our AI Detector".