The AI-generated alt-text on the images in this document is frequently incorrect. For example, the image on page 7 of the bag with human teeth pouring out of it has alt-text that says, "A black and white image of a hat with a red dot on it."
Similarly, the image on page 8 that looks like a disemboweled doll has the alt-text, "A black and white drawing of a doll with a snake for a torso."
The image on page 10 says, "A hand is holding a har with a metal wire around it," when the hand is clearly reaching into the jar, not holding it.
The image on page 12 says, "A woman is standing on a street corner, holding a cell phone." The person does appear to be a woman, and she is holding a cell phone, but there is absolutely no background on this image whatsoever that would suggest a street, much less a corner.
It's also worth mentioning that not all of the images have alt-text.
While I suppose this isn't technically a violation of Paizo's AI policy because it's not "creative work," this very clearly needs better review by an actual human.