Metraipolis › archive › dc_28367fdfc49a

Active Adaptation, Not Static Defense: Temporal Dynamics of Preventative Steering in Adversarial Fine-Tuning — arXiv:2609.10142

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.10142
License (as archived)
CC0-1.0
Archived
2026-10-16T07:26:58.572Z
SHA-256 of the archived text
f96e99939679829a8ee6846d00ce1b3a51816bdfc74038411f1a341fe8df8398
Stable URL of this page
https://metraipolis.com/documento/dc_28367fdfc49a

How to cite

Active Adaptation, Not Static Defense: Temporal Dynamics of Preventative Steering in Adversarial Fine-Tuning — arXiv:2609.10142 — arXiv (en), https://arxiv.org/abs/2609.10142 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_28367fdfc49a (Metraipolis archive dc_28367fdfc49a, sha256 f96e99939679829a8ee6846d00ce1b3a51816bdfc74038411f1a341fe8df8398)
BibTeX
@misc{dc28367fdfc49a,
  title = {Active Adaptation, Not Static Defense: Temporal Dynamics of Preventative Steering in Adversarial Fine-Tuning — arXiv:2609.10142},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.10142},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_28367fdfc49a; sha256 f96e99939679829a8ee6846d00ce1b3a51816bdfc74038411f1a341fe8df8398},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.10142. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large language models remain fragile against malicious fine-tuning, motivating training-time defenses against harmful persona drift. Preventative Steering injects undesirable-trait persona vectors during fine-tuning and removes them at evaluation time, yet the mechanism behind its lasting protection remains unclear. Analyzing its temporal optimization dynamics, we find that the defense emerges from an early compensatory adaptation phase followed by a steady-state phase where the corrective signal decays; in parameter space, attention output projections emerge as the dominant residual-write route for defensive updates. Through Intervention Delta Preservation (IDP) and IDP Continuation experiments, we further show that preserving or reinjecting the weight offset fails to maintain protection, indicating that preventative steering relies on active adaptation rather than a static defense. Motivated by this finding, we propose Progressive Intensity Scheduling (PIS), which starts with a moderate injection strength and increases it after static-strength alignment begins to decay. Across the evaluated Qwen2.5 and Gemma-3 models, PIS improves safety robustness over static-strength steering while reducing harmful trait expression.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-10 in cs.AI).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR); Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.10142 — 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 2026-10-16T07:00:16.391Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-10?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).