Source: https://cybre.space/@tindall/106539167944483388
From the same Mastodon thread:
The model is known to reproduce some code, including GPL-licensed code, verbatim; therefore, it must contain verbatim copies of that code, however it is encoded.
[…]
the snippet in question is clearly, deeply original. it is a cursed coding crime that contains several “magic constants” with high entropy.
So it should be required to be open source now, right?
If the license doesn’t matter to them, they could’ve use Microsoft’s own huge private codebase.