The “abliterators” are flooding the internet with AI models that never say no, making it easier for people to improve their cyberdefenses—or build a bomb.
wired.com/story/tear-down-th…
When you train a cutting-edge AI model, the clever but unruly “base” model goes through additional training called alignment that’s meant to align a model with human ideas about right and wrong.
wired.com/story/tear-down-th…
Many AI experts will tell you that AI must be aligned at all costs. But there’s a community of people who think these fears are way overblown, and that we’re too willingly giving up our liberty in the name of a false sense of safety.
wired.com/story/tear-down-th…
Nov 10, 2026 · 8:06 PM UTC
1
312
Some of these people have started removing the guardrails from popular open-weight models in a process known as abliteration.
wired.com/story/tear-down-th…
1
68
Proponents of abliteration admit that uncensored models can be dangerous, but they say it is fundamentally no different from how a computer or hammer can hurt people if misused. In their eyes, it’s all about putting the user in charge.
wired.com/story/tear-down-th…
1
6,804
This article is part of WIRED’s ongoing series assessing the real threats of artificial intelligence. Read it at the link. 👇
wired.com/story/tear-down-th…
1
6,653