Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
superjose
6 months ago
|
parent
|
context
|
favorite
| on:
GPT-5 leaked system prompt?
I did a search and found reltive terms:
https://www.reddit.com/r/hacking/comments/1kqi0tm/how_canari...
https://medium.com/@tomer2138/how-canaries-stop-prompt-injec...
yencabulator
6 months ago
[–]
Those talk about a mechanism to detect prompt injection. If that had been true, we should have seen the chatbot refuse, not lie.
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
https://medium.com/@tomer2138/how-canaries-stop-prompt-injec...