News for hackers. Stuff that matters.
Re:Open-weight AI is having its Kubernetes moment. Let's not... by holoduke on Saturday July 25, 2026 @04:38PM (#49051328)
All censorship whether it's tienanmen, sex nudity, violence etc are happening in a very thin layer of the model. It's easily bypassed. I use all open models without any censorship. I can easily ask how to make a nuclear bomb or how to build my own meth lab at home with a super details step by step list.
Re:Open-weight AI is having its Kubernetes moment. Let's not... by ButlerianJihad on Saturday July 25, 2026 @04:52PM (#49051446)
Have you followed the steps, though?

Last year I asked some ordinary commercial LLM if it could draw me the schematics of a device like Little Boy from WWII. And it complied very nicely. Got a very detailed schematic graphic. Suitable for framing. Perhaps not quite suitable for reconstruction.

Re:Open-weight AI is having its Kubernetes moment. Let's not... by holoduke on Saturday July 25, 2026 @06:08PM (#49052137)
No I haven't. And I am sure there are flaws in the guide for a nuclear bomb. But a meth lab at home is definitely doable. Would not recommend it