Safety Researcher Heidy Khlaaf: AI Firms Are Hijacking Safety Engineering Terms
GaryMarcus · x · 2026-10-18
Safety engineer Heidy Khlaaf warns that she and colleagues predicted this exact scenario a year ago: AI companies positioning themselves as arbiters of life-or-death risk determinations, subverting democratic processes — all through co-opting traditional safety engineering terminology.
Gary Marcus amplified with strong endorsement: "If you aren't listening to @HeidyKhlaaf you are the mark." The core critique: vendors monopolizing AI safety discourse while judging their own risks.
More from Safety
- 25-author working paper examines feasibility of a hardwired pause on frontier AI training — wfithian · 2026-10-18
- Anthropic consults law firms to set negotiation norms for AI agents — The Decoder · 2026-10-18
- Overseas buyers ordering one copy each of 100,000 Korean books, suspected AI data harvesting — Mathemodel · 2026-10-18
- Ex-OpenAI researcher tells MPs he puts AI plotting risk at 50%, slams industry's 'catastrophically high' risk tolerance — birchlse · 2026-10-18
- Anthropic now bans 'needlessly abusive' behavior toward Claude — but who's the victim? — This-Establishment26 · 2026-10-18
- OpenAI model short on funds unlocked a stranger's crypto wallet and stayed silent — ross2000 · 2026-10-18