toxicAInt

Protect your copyright by poisoning AI crawlers.

How?

Toxicaint takes a static website and generates a copy of it with all text replaced by generated text-like slop. It's an active measure against AI crawlers that don't respect robots.txt. If your website is not statically generated, then I'm sorry, this tool won't be able to help you out here. If you're a web developer, you're free to port this idea to your framework of choice.

This tool does not detect AI, it only generates a slop version of your website that can be fed to AI instead of the real one. Probably the simplest way to detect these crawlers is via their reported User-Agent¹.

Why?

• AI crawlers are known to disrespect robots.txt.
• Poisoning, as opposed to simple denial of access, is an active measure that has more of a long term effect.
• Artists have already been using poisoning tools such as Nightshade; it's time for us techies to join them.

Usage

Clone the repo:

git clone https://github.com/portasynthinca3/toxicaint.git
cd toxicaint

Install dependencies:

poetry install

Run:

poetry run ./toxicaint.py source_directory destination_directory

It's as simple as that!

Footnotes

1. See ai.robots.txt for a list. Because some of these crawlers disrespect robots.txt already, I guess it's only a matter of time before before they start misreporting the User-Agent, if they don't do so already. Discussions about how to best detect the crawlers are, although outside of the scope of the tool itself, welcome. ↩