Post #3977693

@Npars01@mstdn.social @danlyke@researchbuzz.masto.host agreed. And it doesn't have to be done in a clever way, either. LLMs have no "reasoning" built in so you could, for example, make up random sentences involving "AI" companies and known employees of same and the bots would just slurp it up as part of their RAG phase. Not that I'm condoning sabotage or anything... https://www.cia.gov/stories/story/the-art-of-simple-sabotage/