Ask HN: Open-source license that limits usage by ML algorithms?

https://news.ycombinator.com/item?id=33882192

by dakiol • 4 years ago

5 3 4 years ago

news.ycombinator.com

Let’s imagine I want to publish my awesome open source project so that others can contribute, use it for personal usage and all those good things. But I want to explicitly deny usage of my project as training data for any ML algorithm (e.g., copilot, chatGPT, etc.). Does any license exist already to address this? If not, can I just state that as an extra paragraph in my LICENSE file?

nullish_signal• 4 years ago

This seems similar to using robot.txt to block (polite) web crawlers.

Can you "explicitly deny" a reader of your Text/HTML/Code/Image data from learning from it?

tapiok• 4 years ago

you might want to ask here https://forum.artlessdevices.com/

hxugufjfjf• 4 years ago

I asked ChatGPT and got the following answe

It is possible to add a clause to a license stating that the project cannot be used as training data for any ML algorithm. However, it may be more effective to use a license that already includes this restriction, such as the Creative Commons Attribution-NonCommercial-NoDerivatives license. This license allows others to use and share the project for non-commercial purposes, but prohibits the use of the project for any commercial or derivative purposes, including training data for ML algorithms.

I followed up with the following:

where in the Creative Commons Attribution-NonCommercial-NoDerivatives license is it stated the project cant be used for training ml algorithms?

The Creative Commons Attribution-NonCommercial-NoDerivatives license includes a clause stating that the project cannot be used for any commercial purposes, which would include using the project as training data for an ML algorithm that is intended to be sold or used for profit. It also includes a clause prohibiting the creation of derivative works, which could be interpreted as prohibiting the use of the project as training data for an ML algorithm that modifies or builds upon the original project.

Ask HN: Open-source license that limits usage by ML algorithms?

Related Stories