3 ms·
Answering to myself as I can't edit my response any more. I've investigated a bit more and the attack as described by Greshake[1] seems to be much more realisti
by rnosov 4y ago
Answering to myself as I can't edit my response any more. I've investigated a bit more and the attack as described by Greshake[1] seems to be much more realistic to me than I initially thought.
[1] https://github.com/greshake/llm-security https://github.com/greshake/llm-security
- greshake 4y agoYea that's me. It seems to be very difficult right now to get people's attention to this and make them take it seriously. On a side note, your project is also currently putting unfiltered model output straight into osascript sooooo a lot of the fancy gymnastics needed to make stuff work in the paper with only search abilities isn't required in this case.
- rnosov 4y agoJust for the avoidance of doubt "AI Files" is not my project and I'm not affiliated with the author in any way. osascipt line does look a bit dodgy to me but perhaps is safe. But I can see how things might go downhill quickly with this approach...