CTRAP: Embedding Collapse Trap to Safeguard Large Language Models from Harmful Fine-Tuning

TL;DR

This paper introduces CTRAP, a novel method to safeguard large language models from harmful fine-tuning by inducing a controlled collapse that neutralizes malicious adaptations without affecting benign use.

Contribution

The paper proposes a paradigm shift from selective unlearning to inducing model collapse to prevent harmful fine-tuning, with a practical mechanism called CTRAP that activates under malicious updates.

Findings

CTRAP effectively neutralizes harmful fine-tuning across various LLMs.

CTRAP maintains model utility during benign fine-tuning.

Empirical results show high robustness of CTRAP against different attack scenarios.

Abstract

Fine-tuning-as-a-service, while commercially successful for Large Language Model (LLM) providers, exposes models to harmful fine-tuning attacks. As a widely explored defense paradigm against such attacks, unlearning attempts to remove malicious knowledge from LLMs, thereby essentially preventing them from being used to perform malicious tasks. However, we highlight a critical flaw: the powerful general adaptability of LLMs allows them to easily bypass selective unlearning by rapidly relearning or repurposing their capabilities for harmful tasks. To address this fundamental limitation, we propose a paradigm shift: instead of selective removal, we advocate for inducing model collapse--effectively forcing the model to "unlearn everything"--specifically in response to updates characteristic of malicious adaptation. This collapse directly neutralizes the very general capabilities that attackers exploit, tackling the core issue unaddressed by selective unlearning. We introduce the Collapse Trap (CTRAP) as a practical mechanism to implement this concept conditionally. Embedded during alignment, CTRAP pre-configures the model's reaction to subsequent fine-tuning dynamics. If updates during fine-tuning constitute a persistent attempt to reverse safety alignment, the pre-configured trap triggers a progressive degradation of the model's core language modeling abilities, ultimately rendering it inert and useless for the attacker. Crucially, this collapse mechanism remains dormant during benign fine-tuning, ensuring the model's utility and general capabilities are preserved for legitimate users. Extensive empirical results demonstrate that CTRAP effectively counters harmful fine-tuning risks across various LLMs and attack settings, while maintaining high performance in benign scenarios. Our code is available at https://anonymous.4open.science/r/CTRAP.