Using text to train models is legally considered fair use. You can use copyrighted material, you just can't redistribute it. The output is considered the respon

Using text to train models is legally considered fair use. You can use copyrighted material, you just can't redistribute it. The output is considered the responsibility of the user of the software. It's up to the user to ensure code or other content they're using isn't licensed or copyrighted.