- cross-posted to:
- technology@beehaw.org
You must log in or # to comment.
How did they not think about this? This is a very basic prompt injection, and it still falls for it.
They probably asked AI to write the browser. AI loves writing code with security vulnerabilities
The whole attack model has been known for years already and it isn’t even the first time that specifically an LLM browser plugin has been exploited by page contents
https://bsky.app/profile/natanael.bsky.social/post/3kr2ud66y2x24
Why think when there’s VC money to be had?
That’s awesome hahaha
how lovely
