The article’s not titled precisely - it’s specifically about LLM disasters, not AI disasters in general.