LATIDIA · Ciberseguridad

Adaptador de instrucciones de seguridad: un parche de seguridad post-hoc sin capacitación para LLM de peso abierto ajustados

arXiv: 2611.10056v1Tipo de anuncio: nuevo Resumen: Los LLM de peso abierto se ajustan rutinariamente desde repositorios públicos, pero la adaptación específica de la tarea puede erosionar silenciosamente la alineación de seguridad heredada de las bases ajustadas por instrucciones.

La noticia

arXiv:2611.10056v1 Announce Type: new Abstract: Open-weight LLMs are routinely fine-tuned from public repositories, yet task-specific adaptation can silently erode safety alignment inherited from instruct-tuned bases. This is a supply-chain risk when organizations ship third-party checkpoints without full re-alignment. Prior work rarely audits diverse public fine-tunes under a shared multi-axis safety-utility protocol or offers training-free mitigation at deployment time. We propose Safety Direction Adapter (SDA), a training-free post-hoc safety patch that extracts a safety direction once from an aligned base and transfers it one-to-many within that base family as a small PEFT-compatible weight update, evaluated under the same multi-axis protocol. Across six direction extractors, we evaluate 14 checkpoints from two 1-2B