Researcher slams uncensored model makers: massive negative externalities, 'red teaming' as cover
AaronBergman18 · x · 2026-10-22
Aaron Bergman argues that from an economics and policy perspective, the core problem with abliteration-style uncensored models is their massive negative externalities: risk is imposed on the public while providers "cackle and pretend their product is for red teaming." He notes such services deliberately avoid CSAM for legal reasons but allow almost anything else.
More from Safety
- 136-Country, 118M-Article Dataset Audits State Media Share in LLM Training Corpora — deliprao · 2026-10-22
- PHP 8.5.12 may still be vulnerable to php2shell, researchers report — basedjensen · 2026-10-22
- Cloudflare's security-audit-skill hits 35.7k stars, turns coding agents into auditors — Pytlicek · 2026-10-22
- Wiz: stolen credentials from LiteLLM supply chain incident already abused for LLMJacking from two tracked IPs — wunderwuzzi23 · 2026-10-22
- The Supply Chain Dilemma: Patch Fast for Vulnerabilities, Slow for Backdoors — jd_pressman · 2026-10-22
- Security apocalypse is ultimately a test of whether people are morally good — moultano · 2026-10-22