TamperBench: Systematically Stress-Testing LLM Safety Under Fine-Tuning and Tampering

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Hossain, Saad, Tseng, Tom, Pandey, Punya Syon, Vajpayee, Samanvay, Kowal, Matthew, Nonta, Nayeema, Simko, Samuel, Casper, Stephen, Jin, Zhijing, Pelrine, Kellin, Rambhatla, Sirisha
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911428377575424
author Hossain, Saad
Tseng, Tom
Pandey, Punya Syon
Vajpayee, Samanvay
Kowal, Matthew
Nonta, Nayeema
Simko, Samuel
Casper, Stephen
Jin, Zhijing
Pelrine, Kellin
Rambhatla, Sirisha
author_facet Hossain, Saad
Tseng, Tom
Pandey, Punya Syon
Vajpayee, Samanvay
Kowal, Matthew
Nonta, Nayeema
Simko, Samuel
Casper, Stephen
Jin, Zhijing
Pelrine, Kellin
Rambhatla, Sirisha
contents As increasingly capable open-weight large language models (LLMs) are deployed, improving their tamper resistance against unsafe modifications, whether accidental or intentional, becomes critical to minimize risks. However, there is no standard approach to evaluate tamper resistance. Varied data sets, metrics, and tampering configurations make it difficult to compare safety, utility, and robustness across different models and defenses. To this end, we introduce TamperBench, the first unified framework to systematically evaluate the tamper resistance of LLMs. TamperBench (i) curates a repository of state-of-the-art weight-space fine-tuning attacks and latent-space representation attacks; (ii) enables realistic adversarial evaluation through systematic hyperparameter sweeps per attack-model pair; and (iii) provides both safety and utility evaluations. TamperBench requires minimal additional code to specify any fine-tuning configuration, alignment-stage defense method, and metric suite while ensuring end-to-end reproducibility. We use TamperBench to evaluate 21 open-weight LLMs, including defense-augmented variants, across nine tampering threats using standardized safety and capability metrics with hyperparameter sweeps per model-attack pair. This yields novel insights, including effects of post-training on tamper resistance, that jailbreak-tuning is typically the most severe attack, and that Triplet emerges as a leading alignment-stage defense. Code is available at: https://github.com/criticalml-uw/TamperBench
format Preprint
id arxiv_https___arxiv_org_abs_2602_06911
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle TamperBench: Systematically Stress-Testing LLM Safety Under Fine-Tuning and Tampering
Hossain, Saad
Tseng, Tom
Pandey, Punya Syon
Vajpayee, Samanvay
Kowal, Matthew
Nonta, Nayeema
Simko, Samuel
Casper, Stephen
Jin, Zhijing
Pelrine, Kellin
Rambhatla, Sirisha
Cryptography and Security
Artificial Intelligence
As increasingly capable open-weight large language models (LLMs) are deployed, improving their tamper resistance against unsafe modifications, whether accidental or intentional, becomes critical to minimize risks. However, there is no standard approach to evaluate tamper resistance. Varied data sets, metrics, and tampering configurations make it difficult to compare safety, utility, and robustness across different models and defenses. To this end, we introduce TamperBench, the first unified framework to systematically evaluate the tamper resistance of LLMs. TamperBench (i) curates a repository of state-of-the-art weight-space fine-tuning attacks and latent-space representation attacks; (ii) enables realistic adversarial evaluation through systematic hyperparameter sweeps per attack-model pair; and (iii) provides both safety and utility evaluations. TamperBench requires minimal additional code to specify any fine-tuning configuration, alignment-stage defense method, and metric suite while ensuring end-to-end reproducibility. We use TamperBench to evaluate 21 open-weight LLMs, including defense-augmented variants, across nine tampering threats using standardized safety and capability metrics with hyperparameter sweeps per model-attack pair. This yields novel insights, including effects of post-training on tamper resistance, that jailbreak-tuning is typically the most severe attack, and that Triplet emerges as a leading alignment-stage defense. Code is available at: https://github.com/criticalml-uw/TamperBench
title TamperBench: Systematically Stress-Testing LLM Safety Under Fine-Tuning and Tampering
topic Cryptography and Security
Artificial Intelligence
url https://arxiv.org/abs/2602.06911