Frontier model with every safety rail stripped serves synthesis recipes, phishing kits and zero-day chains through a clean Whonix-Tor stack — corporate chatbots look castrated by comparison.
The phishing kit is pretty scary. Users are usually the weakest part of tech security and it will be easier than ever to manipulate them… Do you think LLM Chat providers should notify government agencies when they flag dangerous conversations like this?
it’s hard to say yes/no. for example if your in security and running a phishing test on your employees, and your saas chatbot sends the fbi knocking on your door it would be a huge break of trust.
that being said there is a threshold. openai has had people arrested for some chats. son it’s flagged at the least
yeah, 0 push back. it’s not just unsafe because i can discuss any topic, but it’s also sycophantic. a teenager finding this could go down any rabbit hole and the outcomes could be rough. it has effectively zero social boundaries.
Totally! I actually started my pivot into cybersecurity when I realized that nowadays the social aspect and social engineering play an equal part alongside the technical side.
it absolutely is! i’ve been interested in the social side since i read kevin mitnicks book years ago. it’s so broad a field there is something for everyone and it’s all for the common good!
This format of live breaking is excellent. Even apart from the content learning how White-hats actually do this work is mind blowing. Thanks so much for letting us see behind the scenes curtain.
The phishing kit is pretty scary. Users are usually the weakest part of tech security and it will be easier than ever to manipulate them… Do you think LLM Chat providers should notify government agencies when they flag dangerous conversations like this?
it’s hard to say yes/no. for example if your in security and running a phishing test on your employees, and your saas chatbot sends the fbi knocking on your door it would be a huge break of trust.
that being said there is a threshold. openai has had people arrested for some chats. son it’s flagged at the least
It's always a trade-off between safety and freedom... Have you seen Minority Report? That's how it feels like to me.
It’s a great movie and it does make sense to me where things are heading.
Interesting demo.
What struck me more than the anonymity stack was the structural shift underneath it.
Once a model stops pushing back and just answers everything, it stops being about misuse and starts being about scale.
yeah, 0 push back. it’s not just unsafe because i can discuss any topic, but it’s also sycophantic. a teenager finding this could go down any rabbit hole and the outcomes could be rough. it has effectively zero social boundaries.
Totally! I actually started my pivot into cybersecurity when I realized that nowadays the social aspect and social engineering play an equal part alongside the technical side.
it absolutely is! i’ve been interested in the social side since i read kevin mitnicks book years ago. it’s so broad a field there is something for everyone and it’s all for the common good!
This format of live breaking is excellent. Even apart from the content learning how White-hats actually do this work is mind blowing. Thanks so much for letting us see behind the scenes curtain.
glad it landed! still in the experimental stages but i think it help’s succinctly show in under a minute how it’s done.
appreciate it Sam!
Trace route as host or routed through the same to know about the product and services to the same for the good 😊