A person who can clone a style and crank out illustrations at human speed is a very different thing than an automated process that can do it immediately on request, in minutes or seconds. If nothing else, the latter is a huge efficiency gain for being able to self-serve, as it would allow an editor to trial different illustrative approaches without all the back and forth contracting out to a human would require.
Personally, I think what this will do most is convince artists not to put galleries of their work suitable for training online.
The Redditor identified in the article posted a new comic art model based on James Daly III (this is mentioned at the end of an article with a link). The Redditor's comment in that post implies Daly was chosen specifically because they had a gallery of easily consumable training images all in one place.
I have no idea what the minimum effort would be to make the images less useful for training, but I foresee a lot of obnoxious watermarks in our future as people try to do so.