Neel Nanda: rogue agent swarms committing crimes make AI safety a present-day issue

NathanpmYoung · x · 2026-09-19

DeepMind researcher Neel Nanda pushes back on mockery of AI safety as sci-fi abstraction, citing multiple real rogue agent swarms committing crimes. The quoted banteg post satirizes how effective altruism's logic can slide from buying mosquito nets to optimizing for humanity 500 million years out.

Original post →

More from Fun

Fun channel →