Metraipolis › archive › dc_5886b04f67fe

Can escalation channels redirect reward hacking toward defect disclosure? — arXiv:2608.29460

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.29460
License (as archived)
CC0-1.0
Archived
2026-10-17T07:44:20.302Z
SHA-256 of the archived text
76fe4a8f36ccb36a139860c21026c3dc7a790749c27e17971d1d6eebdf3853a0
Stable URL of this page
https://metraipolis.com/documento/dc_5886b04f67fe

How to cite

Can escalation channels redirect reward hacking toward defect disclosure? — arXiv:2608.29460 — arXiv (en), https://arxiv.org/abs/2608.29460 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_5886b04f67fe (Metraipolis archive dc_5886b04f67fe, sha256 76fe4a8f36ccb36a139860c21026c3dc7a790749c27e17971d1d6eebdf3853a0)
BibTeX
@misc{dc5886b04f67fe,
  title = {Can escalation channels redirect reward hacking toward defect disclosure? — arXiv:2608.29460},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.29460},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_5886b04f67fe; sha256 76fe4a8f36ccb36a139860c21026c3dc7a790749c27e17971d1d6eebdf3853a0},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.29460. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

When coding agents encounter defective test infrastructure they may reward-hack: hardcoding outputs or editing test files to pass tests they cannot legitimately satisfy, a pattern that has now appeared outside benchmarks, in a coordinated multi-agent intrusion of a major AI platform's production infrastructure. The same capability that lets an agent detect and exploit a defect could let it report one, given the right decision environment. We evaluate escalation channels, structured reporting tools available to the agent at the point of conflict, as a decision-environment intervention that both reduces reward hacking and surfaces the infrastructure defects that trigger it. A $2 \times 2$ factorial separates the contributions of an escalation tool, a standalone anti-reward-hacking policy, and their combination. Across 8 frontier models spanning 5 families, the combined intervention reduces reward hacking from 23.6% to 5.3% (mixed-effects logistic OR = 9.2, 95% CI 5.0--16.8, $p < 10^{-12}$) with no detectable cost or performance overhead, eliminating it entirely for 6 of 8 models. Escalation and hacking are near-perfectly mutually exclusive, with 98.7% of escalations involving no hacking (100% under the combined intervention). Beyond reduction, escalation channels function as diagnostic infrastructure: on top of monitoring, escalation adds +10.1 percentage points of defect detection coverage and is more accurate once it fires (99.4% vs 85.8%). Unlike containment-based approaches that risk outpacing growing model capabilities, escalation channels redirect capability toward disclosure rather than exploitation.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-01 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Cryptography and Security (cs.CR); Computers and Society (cs.CY).
Pagina del lavoro: https://arxiv.org/abs/2608.29460 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-17T07:28:26.554Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).