I wrote a program (FlashQard) that was discontinued about 15 years ago. Recently, I received some emails about new comments on its opensuse packages which at first seemed genuine and thought maybe somebody is still interested. But, it appears the comments were most likely generated by a language model, only rephrases the package description, and a completely irrelevant link was injected in the comments. I checked out a few other packages and they also have similar comments.
Have you seen something like it? I wonder how far the use of language models will go in regards to spams and if/when HN will be targeted.