Characterizing and Defending Utility-Hijacking in Instruction-Tuning Data Selection

Yahao Ren, Jian Lou, Huadi Zheng, Jiawei Duan · 2026 · arXiv

DOI: 10.48550/arXiv.2610.29886 · open-access PDF

Abstract

Instruction-data selectors choose which instruction-following records are used for fine-tuning, improving alignment quality, raising training efficiency, and filtering out harmful records. A selector's choices thus directly affect the fine-tuned model's safety, yet despite their widespread use, the security of these selectors remains under-studied. We conduct the first systematic study of adaptive supplier-side attacks against utility-oriented instruction-data selectors and introduce utility-hijacking. Utility-hijacking shapes harmful records to remain compatible with a selector while preservi

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Is this paper real? This page was built from OpenAlex and arXiv records, not generated. Software can verify any reference in one call: GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.29886, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txt

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange