AIGizmodoabout 2 hours ago
Abliteration Launches Unsafeguarded AI Model
1 min read
Abliteration, a Palo Alto startup, launched abliterated-model-large-v2 on Monday, built on Z.ai's GLM-5.3 with safeguards removed via orthogonalization. The model handles offensive cyber and red teaming tasks other AI refuses, but won't generate CSAM or self-harm text.
Level
Hype check
Tap to vote and see what everyone thinks.
Summary by ByteBrief
