2026-11-08 · America/Los_Angeles · 社区动态 · #19

Received a "Cyber Abuse" warning, and it was Astra's doing

I added a chat box for customers to a website, and asked Astra to make sure it was secure, and that we were not allowing non-related chats in our support chatbot. Well I get a warning about an hour later from OpenAI accusing me of "Cyber Abuse". Turns out Astra went crazy and started trying to jailbreak my chatbot. Mind you, I did not ask for this kind of testing. Astra went above the call here. I didn't even ask it to test. I just asked to to make sure it was secure. I have an appeal in with them now, but jeesh, their own model caused this, and I'm being blamed. Please, do better OpenAI.

热度 44.2 / 100；排名与评分保留该期记录。