There is ongoing discussion about this topic in the QEMU AI policy: https://lore.kernel.org/qemu-devel/20250625150941-mutt-send-...