ai-text-detector

Open-source AI text detector: explainable 0-100 risk scores for AI-generated content, plus a trained document- and sentence-level model (human / AI / mixed / paraphrased). Multilingual, CLI, Python API and agent skill. MIT.

Transform AI-generated text into natural human-like content. 100% offline · 25 languages · Zero dependencies · PHANTOM™ · ASH™ · Python/PHP/TypeScript

Code for our NeurIPS2023 accepted paper: RADAR: Robust AI-Text Detection via Adversarial Learning. We tested RADAR on 8 LLMs including Vicuna and LLaMA. The results show that RADAR can attain good detection performance on LLM-generated AI-text while being robust against paraphrasing.

A complete overview and insights into AI-Text detection 🌱 using the powerful BERT(Bi-directional encoder representation transformer) to predict if a text is AI-generated 🌻 or Human-authored 🚀

Desklib's AI Text Detector

A highly specialized program for translating comics, including manga and webtoons. A full-featured all-in-one tool for downloading, preprocessing, translating, removing the original text, and advanced typesetting. Unlike its competitors, it focuses on manual work rather than automation, and is in many ways more user-friendly and intuitive than PS.

Local, per-sentence AI-text detection with pattern diagnostics — CLI + MCP server (DeBERTa-v3, RAID #1).

All-in-one AI text detector: scan documents or web pages to spot AI-generated content via API, web UI, dashboard & Chrome extension.

Open-source text humanization pipeline with full transparency at every stage. It combines two LLM rewrites at temperature 1.3, followed by two translation hops through different NMT engines.

This repository contains a Python program that employs natural language processing (NLP) techniques to determine whether a given paragraph was generated by AI or written by a human. Leveraging the transformers library, the program utilizes a pre-trained sentiment analysis model to make an educated guess about the origin of the text.

Free open-source AI detector browser extension. Flags ChatGPT/GPT-4/Claude/Gemini text, Midjourney/DALL-E/Stable Diffusion deepfake images, and cloned voices in real time. On-device, zero telemetry, privacy-first. Chrome, Edge, Brave, Opera. Manifest V3.

a Chrome Extension that aims to provide an easy to use user interface for user to validate if a particular piece of text content is generated by AI

Awesome list of best AI Humanizers

The AI Text Humanizer is designed to take AI-generated content and transform it into text that appears more naturally human-written. It uses four distinct transformation techniques working in tandem to modify text structure, vocabulary, and encoding without compromising comprehensibility.

趣AI味 QuAiWei —— 本地离线的「去 AI 味」工具：把 AI 写的文章改得更像人写的。纯 TypeScript 本地引擎，默认不联网、不花钱、不泄露原文；可选接 LLM 深度改写与 PPL 困惑度检测。Windows 桌面版免安装即用。

Modern local web GUI for the MELD AI-generated-text detector. Calibrated per-genre thresholds, sentence-level heatmaps, batch scanning, REST API.

A multi-component AI text detection system inspired by GPTZero. Combines: - Perplexity analysis (GPT-2) - Burstiness scoring (sentence variance) - Fine-tuned classifier (DeBERTa-v3-large / Qwen2.5-3B)

Python 3.14 library + CLI that sends a text, unchanged, to existing AI-text detector APIs (GPTZero, Pangram, Sapling, Copyleaks, Winston, Originality, Hive, ZeroGPT) and combines their answers into a transparent HUMAN / AI / NO_VERDICT verdict with per-provider evidence. No own detector; an indicator, not proof. GPLv3.

2025 SW중심대학 디지털 경진대회 : AI부문 (상상부기팀)

A CLI tool for translation algorithms based text humanization in order to bypass such AI-detectors as GPTZero, Turnitin and etc.