Inside the 'abliterators': the renegades stripping guardrails from open-weight AI models
willknight · x · 2026-11-11
A WIRED deep dive into the growing "abliteration" community—people who remove safety guardrails from open-weight models so they never say no.
- Former Apple engineer Devon Thomas founded Abliteration AI, which strips safeguards from the most powerful open-weight models and offers them to anyone.
- His motivation: frustration with hosted models' privacy risks, unpredictable behavior changes, and over-refusals.
- Notable jailbreak researcher Pliny frames the movement as a fight for "cognitive freedom," comparing it to the Cypherpunks.
The piece challenges the assumption that uncensored models serve mainly criminals—showing both defensive upsides and dangerous downsides of the guardrail-free ecosystem.
More from Safety
- Microsoft research: editing fewer than 10 neurons backdoors any AI agent — DimitrisPapail · 2026-11-11
- Polymarket odds: 8% chance Anthropic fully pauses AI training by end of year — Polymarket · 2026-11-11
- User fakes hostage crisis to trick Google AI into recommending cracked-software sites — SumitGup · 2026-11-11
- OpenAI DevDay 2026: Codex Security goes from vulnerability to verified fix — OpenAI · 2026-11-11
- VulnCheck audit: Anthropic's Glasswing logged 442 CVEs but only 217 fixes in 7 months — jedisct1 · 2026-11-11
- UC Berkeley's SubWM watermark detects LLM distillation even after laundering attacks — sumanthd17 · 2026-11-11