Now imagine if Russia or China dedicated some resources for such... training set data flaws based attacks? Other attacks might also be worth a go, like trying to bait ChatGPT to recreate some works like Lolita or freaky "loli" fanfic sites, and if it succeeds pushing a PedoGPT narrative on social media.

Hopefully those companies took it into consideration, right? Seems like a very fucking obvious vector of attack.