Open-model advocates push back on blames that abliterated LLMs drive vulnerability surge
max_paperclips · x · 2026-10-23
Deedy argued that part of the surge in critical vulnerabilities stems from 'abliterated' models with safeguards removed — citing a guardrail-stripped GLM 5.3 available for $10 as a world-class pentesting tool.
Dan Jeffries pushed back hard, calling the narrative an agenda-driven info war against open models: the PHP vulnerabilities were found by Sol 5.6 during a defensive audit, not by abliterated models. 'If you have magic glasses that show which buildings will fall, you fix the buildings, not smash the glasses.' He added that when HuggingFace and OpenCode were under attack, closed models refused to help and open models did the defense — and while OpenCode's prominence got it access to OpenAI's Daybreak program, individual researchers are routinely denied with obscure, unexplained errors.
More from Models
- Dev plans to point an abliterated local model at his smart TV to strip spyware — TheZachMueller · 2026-10-23
- Claude $200 plan delivers 45.3x API value, author denies Anthropic quota cut — RexDouglass · 2026-10-23
- User slams OpenAI's GPT-6.1 Sol: 20 hours lost, ignores human feedback — RexDouglass · 2026-10-23
- Dexmal's open-source VLA model DM0.5 lands in LeRobot, tops RoboColiseum — AdinaYakup · 2026-10-23
- Gemini tip: enable agentic video understanding for ~90% lower token usage — patloeber · 2026-10-23
- OpenAI's Tibo Sottiaux confirms a Codex reset is landing tomorrow amid quota complaints — jeppa35 · 2026-10-23