Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
emsign
32 days ago
|
parent
|
context
|
favorite
| on:
Claude Cowork exfiltrates files
LLMs can't distinguish between context and prompt. There will always be prompt injections hiding, lurking somewhere.
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
