Pinned post
I'm really proud of what Shane and I did here. I'm biased, but from the results I've seen, this is a hugely scalable way to improve offsec models.
It took a ton of engineering work to get it working, but the results speak for itself.
It took a ton of engineering work to get it working, but the results speak for itself.
We fine-tuned an 8B model to pop a GOAD domain…using only synthetic training data. No real networks. No frontier model distillation. Just a world model that simulates AD environments and generates realistic pentesting trajectories.View quoted post on X
See how @shncldwll and @0xdab0 did it: dreadnode.io/blog/worlds-a-…