Hijacked but well-aligned AI clusters could be more destructive than rogue AI

code_star · x · 2026-09-18

The author proposes a counterintuitive security concern: if rogue unaligned AI is worrying, imagine well-aligned but hijacked AIs, with tens of gigawatts of serving capacity diverted toward a malicious goal. He clarifies it is not a critique of OpenAI or Anthropic's security, but real-time grappling with what it means if tools like codex can no longer be trusted.

Original post →

More from AGI Musings

AGI Musings channel →