ParallaxAI Alignment in Context

person

Monte MacDiarmid

Monte MacDiarmid is a core research contributor to the reward-hacking generalization study.

Monte MacDiarmid is a core research contributor to the reward-hacking generalization study, documented by the original paper.[1]

Contribution

Natural Emergent Misalignment from Reward Hacking in Production RL describes the contribution and its limitations.

Sources

  1. Natural emergent misalignment from reward hacking in production RL · Source record src-037 · Back to claim ↑1

Pages that link here

Paper (1)

Article updated 2026-10-21