Ask HN: State of permissive licences and AI training

1 points by sixthDot ↗ HN
We, the many, share code with permissive licenses. A few years ago, when GitHub released copilot, the issue was raised. Nowadays it's even worse.

Do we have, 05/04/2024, a solution in term of open source license ? A way to guarantee that the code we share will be only used by humans ?

3 comments

[ 3.0 ms ] story [ 28.2 ms ] thread
No guarantee but you can add legal language in the license text specific to your code associated with the license that restricts use of the OSS for training models, using code in running models, etc.
what's the most recognized definition of "human being" ? Does the Cambridge dictionary make authority ?

Also the fact to automate an usage of the code may be difficult to define, for example a script that build VS a script that scraps.

I do not have the knowledge to write a legal statement. That's a bit why I ask, in first place.

I'm on this right now

---

The content of the files included in this reposity can only be consulted by inherited human vision.

Visually unpaired people can obtain the right to use indirect systems to be informed of the content by explicit demand to the author.

---

1. You can read if you're a human

2. If you cant read you might be a human but you have to ask.