A plagiarism detector's natural predator, written in pure regex

What it does

Turnitout is a rule-based text processor that rewrites documents, LaTeX, DOCX, and source-code comments to evade plagiarism and AI-detection tools (Turnitin, GPTZero, CopyLeaks, ZeroGPT). It also scrubs invisible zero-width Unicode characters, claims to disrupt cryptographic watermarks like SynthID, and strips C2PA metadata from files. It runs as a Streamlit web app with several mirror instances, plus bulk ZIP processing and a built-in citation generator that produces BibTeX and LaTeX references across five academic fields.

The interesting bit

The pitch is that it uses no generative AI at all — just deterministic rules that break up n-gram sequences detectors rely on, while supposedly leaving equations, code logic, and document formatting intact. That’s the opposite of commercial “humanizers,” which rewrite with an LLM and arguably leave a fresher AI footprint behind.

Key highlights

• Section-aware processing: rewrites prose but locks down code blocks, data, and declarations
• Comment-only rewriting for Python, C/C++, Java, JS, Go, Rust, SQL, and more, with beginner-to-expert tone personas
• Zero-width character purge (U+200B and friends) and C2PA metadata stripping
• Bulk ZIP in/out with folder hierarchy preserved; claims in-memory-only processing with no storage
• Auto-citation generator exporting BibTeX, \cite{} keys, and JSON

Caveats

• The claims are extraordinary and entirely self-asserted — no benchmarks, no third-party verification, and the README’s tone (“I love hurting billion-dollar companies, hehehe”) doesn’t inspire audit-grade confidence
• Its stated purpose is defeating academic integrity systems, which is a fast way to violate your institution’s policies regardless of whether it works
• “Shattering” statistical watermarks like SynthID with regex-level rules is a strong claim the README never substantiates

Verdict

As a technical curiosity — deterministic n-gram disruption, Unicode sanitization, comment-aware code rewriting — it’s genuinely interesting. As an academic tool, it’s a cheating aid with unverified marketing; read it for the techniques, not the promises.