AI models released under open source license without original training data or program are not seen as DFSG-compliant.

The GR is in his discussion step, and may be other vote options are presented.

The discussion can be read in the Debian-vote mailing list, and the resolution could impact other software (scanners, text2speach…) ^[1].