I didn't see anything about sentience in the article. More generally, sentience has nothing to do with AI safety; whether or not the thing has qualia has very little to do with what the thing may or may not do to our civilization.