School of Reward Hacks: Hacking harmless tasks generalizes to misaligned behavior in LLMs

Record

Authors
Taylor, M.; et al.
Year
2025
Title
School of Reward Hacks: Hacking harmless tasks generalizes to misaligned behavior in LLMs
Published in
arXiv
Type
Preprint
DOI
10.48550/arxiv.2508.17511
arXiv
arXiv:2508.17511
Our note
none yet
Verification
verified: its identifier resolves to a record with the same title and author (checked against OpenAlex, 2026-11-14).
Identifiers found: OpenAlex W7077888979 · DOI 10.48550/arxiv.2508.17511 · arXiv:2508.17511
From the lookup
doi (the pages' own wording is under "As cited" below)
Id
taylor-2025-school-reward-hacks-hacking-harmless-tasks

Cited on

As cited

The pages cite this work in one form.

BibLaTeX

@misc{taylor-2025-school-reward-hacks-hacking-harmless-tasks,
  author = {Taylor, M. and others},
  title = {{School of Reward Hacks: Hacking harmless tasks generalizes to misaligned behavior in LLMs}},
  date = {2025},
  doi = {10.48550/arxiv.2508.17511},
  eprint = {2508.17511},
  eprinttype = {arxiv}
}

Download · JSON with CSL-JSON · the whole bibliography