# OmniDetect

> Know what to fix before you submit. OmniDetect shows which passages of a document read as AI, how strongly, and what to change — then lets you re-check. Detection is an estimate, so we publish how our own measurement behaves, misses included.

OmniDetect helps students, educators, writers and content teams answer one practical question: will this document trigger an AI flag, and if so, what exactly do I do about it? A single score never answers that. We show the passages that drew attention, the strength of each signal, and the limits of what any detector can say.

We operate a detection system, not a detector brand: pluggable analysis channels (commercial engines and LLM-based readers) sit behind our own measurement, routing and evidence layer. Channel names are disclosed as signal sources in reports, not sold as a lineup.

## What you get
- Pre-submission check for text (Japanese and English), with passage-level evidence and known limits
- Free guest scans daily; registered free quota; a one-time paid complete check with a free re-check after edits
- Targeted rewrite suggestions for confidently flagged passages — reviewed and re-checked by you (we do not promise to bypass any detector)
- AI code detection (English): paste a file, get an honest reading with its measured limits

## Core Pages
- [Homepage](https://www.omnidetect.ai/): Free check — know what to fix before you submit
- [AI Code Detector](https://www.omnidetect.ai/ai-code-detector): Free code check with measured limits and demo samples
- [Accuracy & Methodology](https://www.omnidetect.ai/accuracy): How we measure, by language and register
- [Pricing](https://www.omnidetect.ai/pricing): Free tier, one-time complete check
- [How It Works](https://www.omnidetect.ai/how-it-works): The detection pipeline and its boundaries

## Measurement & Evidence
- [Japanese Detection Benchmark](https://www.omnidetect.ai/ja/benchmark): 221 frozen documents, six detectors, one public matrix — including our own false positives (5 of 25 student-register documents, currently)
- [Theory: how detectors work (JA)](https://www.omnidetect.ai/ja/learn/how-detectors-work)
- [Theory: reading accuracy claims (EN)](https://www.omnidetect.ai/learn/how-to-read-accuracy-claims)
- [Theory: why reports don't prove authorship (EN)](https://www.omnidetect.ai/learn/why-reports-dont-prove-authorship)
- [Detector Disagreement Index](https://www.omnidetect.ai/guide/detector-disagreement-index): How often independent detectors contradict each other on the same text

## Task Pages (English)
- [Best AI Detector](https://www.omnidetect.ai/check/best-ai-detector)
- [Free AI Detector](https://www.omnidetect.ai/check/free-ai-detector)
- [AI Essay Checker](https://www.omnidetect.ai/check/ai-essay-checker)
- [Detect ChatGPT](https://www.omnidetect.ai/check/detect-chatgpt)

## Japanese (日本語)
- [ホーム](https://www.omnidetect.ai/ja): 提出前にAI判定の反応を確認
- [自分で書いたのにAI判定されたとき](https://www.omnidetect.ai/ja/guide/gokenchi)
- [AI検出の理論と限界(シリーズ)](https://www.omnidetect.ai/ja/learn/ai-detection-limits)

## Other Locales
- [Deutsch](https://www.omnidetect.ai/de) · [Français](https://www.omnidetect.ai/fr) · [Português](https://www.omnidetect.ai/pt) · [한국어](https://www.omnidetect.ai/ko) · [Español](https://www.omnidetect.ai/es) · [Nederlands](https://www.omnidetect.ai/nl)

## Privacy
Submitted documents stay in the owner's history and can be deleted at any time, account or not. We do not sell submissions; the only outside parties that process text are the analysis channels named in our privacy policy. Detection is an estimate and is never proof of authorship or misconduct.
- [Privacy Policy](https://www.omnidetect.ai/privacy)
- [Terms](https://www.omnidetect.ai/terms)