AI Matters
Why it matters
New service launches to strip safety guardrails from AI models
The Decoder Stripping safety guardrails from open-weight AI models is now a turnkey commercial service
Startup Abliteration.ai launched a commercial service that removes safety mechanisms from open-weight AI models, which researchers used to easily generate malware.
Why it matters
The commercial availability of safety-stripped models makes it easier for bad actors to generate malware and bypass standard AI safety restrictions.
Latest brief · Oct 19, 2026