2 ms·
OpenAI has blocked numerous jailbreaks (despite claiming their model is unchanged). How hard would it be for them to plug this. Also, what’s the nature of this
by cornercasechase 3y ago
OpenAI has blocked numerous jailbreaks (despite claiming their model is unchanged). How hard would it be for them to plug this. Also, what’s the nature of this attack? It’s really unspecific in the article.
- trolan 3y agoThe model itself was fine tuned for JSON function responses, they admitted that openly. They also acknowledge they make changes to ChatGPT all the time, which has nothing to do with the model underneath it.