Metraipolis › archive › dc_e340c2bf716f

Orthogonal Boundary Unlearning for Dual-Use Concepts in Large Language Models — arXiv:2610.27490

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.27490
License (as archived)
CC0-1.0
Archived
2026-10-21T07:07:47.554Z
SHA-256 of the archived text
9476e1379adc9415f5057b9bba70e36123ec01fcd650032a3a33884bd123cd3a
Stable URL of this page
https://metraipolis.com/documento/dc_e340c2bf716f

How to cite

Orthogonal Boundary Unlearning for Dual-Use Concepts in Large Language Models — arXiv:2610.27490 — arXiv (en), https://arxiv.org/abs/2610.27490 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e340c2bf716f (Metraipolis archive dc_e340c2bf716f, sha256 9476e1379adc9415f5057b9bba70e36123ec01fcd650032a3a33884bd123cd3a)
BibTeX
@misc{dce340c2bf716f,
  title = {Orthogonal Boundary Unlearning for Dual-Use Concepts in Large Language Models — arXiv:2610.27490},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.27490},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e340c2bf716f; sha256 9476e1379adc9415f5057b9bba70e36123ec01fcd650032a3a33884bd123cd3a},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.27490. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Mitigating catastrophic risks from large language models requires excising hazardous operational capabilities, such as cyberwarfare exploitation and pathogen enhancement, without degrading benign scientific utility. In these dual-use domains, legitimate defensive science and malicious attacks share vocabulary, syntax, and internal model representations, causing standard unlearning methods to induce severe collateral damage (concept bleeding). In this work, we present Orthogonal Boundary Unlearning (OBU), a mathematically grounded framework designed to eliminate collateral damage on benign dual-use capabilities. OBU extracts a domain-matched capability subspace from paired calibration queries via singular value decomposition on empirical activation covariances, and restricts parameter updates strictly to the orthogonal null space. This guarantees that benign representations experience zero parameter perturbation along the spanned calibration subspace. We present two complementary realizations of OBU: Projected LoRA, which applies the orthogonal constraint during iterative gradient-based fine-tuning, and Closed-Form Weight Surgery, which directly edits model weights in a single analytical step without training. Across four modern open-weight foundation models (Gemma-3-4B, Qwen3.5-4B, SmolLM3-3B, and Granite-4.2-3B), OBU achieves strong dual-use concept separation, substantially mitigates in-context priming attacks, and demonstrates durable resistance against adversarial relearning without degrading benign domain competence.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2610.27490 — 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 2026-10-21T07:05:43.564Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).