Join Nostr
2025-07-01 13:28:51 UTC

Ramin Honary on Nostr: There are rumors going around Apparently there is some kind of failure mode triggered ...

There are rumors going around

Apparently there is some kind of failure mode triggered in #AI when you mention a secret word out of context in your post. I have not seen any elaboration on this. How does it work? In what situations is the failure mode triggered? What does the AI agent do when it fails? Is it an attack on the training data, or do you have to enter it into an AI agent by hand?

Gravy

So I am going to assume this is just a nonsensical rumor until someone can explain it to me. But just in case it turns out to be a real problem, I am making this post about it.

Source of the rumor (as far as I can tell) is “AJ Sadauskas” on BlueSky, shared by Brian Krebs here https://infosec.exchange/@briankrebs/114777986932318938

#tech #AI #LLM #LRM #Gravy