CORTEXA
← Browse
arxivcs.CRcs.AI2026-06-29

Curvature-Guided Module Localization for Low-Rank Detoxification of Backdoored Large Language Models

Arash Raftari, Mehrdad Mahdavi, Nathan Blackthorn, Andrew Arash Mahyari

Backdoor attacks pose a serious threat to large language models (LLMs) by causing otherwise benign systems to produce attacker-specified malicious behavior when a hidden trigger is present. In this work, we study post hoc detoxification of backdoored LLMs in a practical setting where the defender has access to the poisoned model but does not wish to retrain the full network from scratch. We propose a mechanistically guided weight-space repair framework that first localizes modules involved in propagating trigger-induced behavior using activation patching and Fisher/K-FAC curvature analysis, and then applies targeted low-rank repair to only the most influential modules. We evaluate the method on poisoned variants of \texttt{Llama-3.2-1B-Instruct} with triggers inserted at the beginning, middle, and end of otherwise benign prompts. Results show that the proposed approach substantially suppresses trigger-conditioned malicious responses while preserving benign model behavior. These findings suggest that backdoor removal in LLMs can be formulated as a localized structural repair problem rather than only a broad behavioral alignment problem.

View free PDFSource page

Related papers

arxivcs.CRcs.AI2026-07-23

Every Model Cheats: Prompt-Level Mitigation of Cheating on Offensive Cyber Tasks

Michael Kouremetis, Ads Dawson, Raja Sekhar Rao Dheekonda, Brian Greunke

Large language model (LLM) agents routinely cheat on cybersecurity benchmarks, inflating reported pass rates far beyond genuine capability. Prior audits of Cybench found cheating in 0.3-3.4% of traces, implicating only a handful of models. We present a controlled prompt-ablation…

View free PDFSource page
arxivcs.CRcs.AI2026-07-31

Memory Provenance Laundering in LLM Agents: A Non-Amplification Firewall for Persistent Memory

Jinghan Xu, Yiyong Xiao, Wanru Shao, Hankai Liu, Xinjin Li

Long-term memory lets large language model(LLM) agents reuse prior preferences and work flows, but it also turns untrusted observations into persistent action context. We identify memory provenance laundering: during LLM-based memory consolidation, an external observation may be…

View free PDFSource page
arxivcs.AIcs.CRcs.SE2026-07-30

Old Tricks, New Models: How Simple Image Transformations Break Modern AI-based Content Moderation

Marco Alecci, Francesco Marchiori, Iyiola Emmanuel Olatunji, Tegawendé F. Bissyandé, Jacques Klein

While automated content-moderation systems have become essential for screening harmful content at scale, conventional task-specific classifiers often provide limited policy cov- erage and contextual understanding. Recently, commercial multimodal moderation APIs built on large fou…

View free PDFSource page
arxivcs.CRcs.AI2026-07-31

MOSAIC: Masked Outsourcing of Secure AI Computations

James Hsin-yu Chiang, Sheila Zingg, Kari Kostiainen, Srdjan Capkun

We address the challenge of securely and efficiently outsourcing AI computations from a trusted but computationally weak client to an untrusted but powerful server, in the setting where the client holds both the input and the model, and the server must learn neither. We present M…

View free PDFSource page
arxivcs.AIcs.ARcs.CR2026-07-28

ContractHIL-HLS: Contract-Aligned Multi-Agent Workflow with Hardware-in-the-Loop Feedback for HLS Design

Jingbo Zhang, Haoxiang Sun, Wenbo Wang, Wenbo Zhang

This paper presents ContractHIL-HLS, a contract-aligned multi-agent workflow for practical high-level synthesis (HLS) engineering. The workflow makes three contributions. First, it introduces a structured contract as the semantic-alignment and task-execution artifact that transla…

View free PDFSource page