Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
jeremyjh
29 days ago
|
parent
|
context
|
favorite
| on:
Agent Safehouse – macOS-native sandboxing for loca...
Prompt injection attacks are very much a thing. It doesn't matter how good the agent is, its vulnerable, and you don't know what you don't know.
ramoz
29 days ago
[–]
Where are we at with SOTA or reliable prompt injection detection mechanisms?
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search: