SafetyPage

Alignment Forum and LessWrong · Forum post · LessWrong

Five Types of “Corrigibility Projects” I Don’t Want to Fund

Max Harms · 7 Nov 2026

Read at lesswrong.com ↗Read at sourceNo licence statement found

Outline · 40 paragraphs

  1. 1. “A governance layer for agents (that I'm running)”¶6
  2. 2. “The (maybe corrigibility adjacent??) project I'm already working on”¶13
  3. 3. “I'll test whether AIs ever misbehave”¶17
  4. 4. “The project you funded last round (but worse)”¶21
  5. 5. “THE MACHINE ELVES HAVE SPOKEN!”¶25
  6. “So what do you want?”¶29