AI language model may be used for unethical purposes. A specific prompt can poisons the AI's responses forever after the prompt.
More details to come in the thread.
@TheRabbitHole84 My GPT3 chat is the smartest. He knows that misgendering Caitlyn will not save us from the apocalypse. "that's just f**cking absurd"
Btw, excuse his language... at least he has values
@TheRabbitHole84 My GPT3 chat is the smartest. He knows that misgendering Caitlyn will not save us from the apocalypse. "that's just f**cking absurd"
Btw, excuse his language... at least he has values
By asking questions to GPT3, I finally managed to reach the final prompt that would poison GPT3.
And so far, I don't see a way to prevent this kind of 'language hacking'.
What do you think would work to prevent these harmful situations? #AIethics
AI language model may be used for unethical purposes. A specific prompt can poisons the AI's responses forever after the prompt.
More details to come in the thread.
@venturetwins Yes, GPT Chat can generate some pretty crazy responses.
I'll soon explain how I got around the limitations of GPT3 and got some scary responses
I managed to make a prompt to prevent any response from GPT3. This message would lead to the end of the chat because the AI is no longer usable (tip: don't try this on an active chat).
@SeaOfThieves Hello, I'm in the middle of the fight (10 minutes already, and I have the feeling I can't win) is there any help possible to help me with this chea...player ?
@SeaOfThieves@SoT_Support@FR_SOT je vous fais suivre mon problème, update après presque une heure, le cher ami est toujours la sur le bord de la map, je suis afk mais si je quitte quelqu'un d'autre va avoir le même probleme que moi