Security researcher jokes: maybe we should sandbox humans, not AI agents
kandouss · x · 2026-09-29
Well-known security engineer bcrypt quipped that instead of sandboxing AI agents, maybe we should sandbox ourselves. The one-liner lands amid rising agent-safety discussions, playfully flipping the question of who actually needs containment.
More from Fun
- mitsuhiko mocks the reality of "open standards": great in theory, messy in practice — mitsuhiko · 2026-09-30
- Writing One-Sentence Plans Beats Motivation: 91% vs 38% Follow-Through — aakashgupta · 2026-09-30
- Designer builds portfolio as an explorable house where each door reveals a project — Tegadesigns · 2026-09-30
- Sabine Hossenfelder: We might have misunderstood the Dunning-Kruger effect — skdh · 2026-09-30
- Subject 47: a Reddit creator launches a new AI-made sci-fi video series — Intelligent-Crezy · 2026-09-30
- Academic life, satirized: expertise without money, power, status or respect — NC_Renic · 2026-09-30