Abstract
Mechanistic edits (ablations, weight edits, activation steering) are the standard tools for unlearning a harmful capability from a neural network while preserving useful ones. Current approaches validate their effects only by testing, which can never cover an entire continuous region of inputs. Prior work at the interpretability-verification boundary certifies descriptions of a model: what a circuit computes, or whether it faithfully explains the whole. We instead certify the behavioral effect of an edit: that disabling a circuit removes one skill and provably preserves another, for every input in a region; a feature non-interference guarantee in the information-flow-security sense. We demonstrate such certified edits from toy ReLU networks up to a standard softmax + LayerNorm transformer, proving removal and preservation over continuous embedding-space regions and reaching roughly 9x the input-perturbation dimension an exact solver can handle by switching to sound bound propagation. Furthermore, we prove that no finite deterministic black-box test can certify removal, exhibiting an edit that passes exhaustive testing yet provably fails on a survivor pocket that can be made arbitrarily small. Guarantees hold on small, standard-architecture networks and, like any removal claim, presuppose that the target skill admits a decidable specification, a property which real-world harms may not have.
Keywords
Subject
Publication details
- Journal
- Not available
- Open access
- Green open access
Cite this article
APA 7
Uddin, M. S., Mazumder, M. K. A., & Mridha, M. F. (2026). Certified Mechanistic Edits: Behavioral Guarantees for Skill Removal and Preservation. https://omanscience.com/en/articles/certified-mechanistic-edits-behavioral-guarantees-for-skill-removal-and-preservation
MLA 9
Uddin, Md Sazid, et al. "Certified Mechanistic Edits: Behavioral Guarantees for Skill Removal and Preservation." https://omanscience.com/en/articles/certified-mechanistic-edits-behavioral-guarantees-for-skill-removal-and-preservation.
Chicago (author–date)
Uddin, Md Sazid, Md. Khairul Alam Mazumder, and M. F. Mridha. 2026. "Certified Mechanistic Edits: Behavioral Guarantees for Skill Removal and Preservation." https://omanscience.com/en/articles/certified-mechanistic-edits-behavioral-guarantees-for-skill-removal-and-preservation.
Harvard
Uddin, M. S., Mazumder, M. K. A. and Mridha, M. F. (2026) 'Certified Mechanistic Edits: Behavioral Guarantees for Skill Removal and Preservation', Available at: https://omanscience.com/en/articles/certified-mechanistic-edits-behavioral-guarantees-for-skill-removal-and-preservation.
Vancouver
Uddin MS, Mazumder MKA, Mridha MF. Certified Mechanistic Edits: Behavioral Guarantees for Skill Removal and Preservation. https://omanscience.com/en/articles/certified-mechanistic-edits-behavioral-guarantees-for-skill-removal-and-preservation
IEEE
M. S. Uddin, M. K. A. Mazumder, and M. F. Mridha, "Certified Mechanistic Edits: Behavioral Guarantees for Skill Removal and Preservation," https://omanscience.com/en/articles/certified-mechanistic-edits-behavioral-guarantees-for-skill-removal-and-preservation.