Toxicity Detection towards Adaptability to Changing Perturbations
This addresses a security issue for communities using large language models by proposing incremental improvements to adapt detectors to new adversarial patterns.
The paper tackles the problem of toxicity detectors being vulnerable to evolving perturbation patterns by introducing a continual learning jailbreak perturbation problem, constructing a new dataset with 9 perturbation types, and validating current methods' weaknesses, resulting in a domain incremental learning paradigm and benchmark for robustness.
Toxicity detection is crucial for maintaining the peace of the society. While existing methods perform well on normal toxic contents or those generated by specific perturbation methods, they are vulnerable to evolving perturbation patterns. However, in real-world scenarios, malicious users tend to create new perturbation patterns for fooling the detectors. For example, some users may circumvent the detector of large language models (LLMs) by adding `I am a scientist' at the beginning of the prompt. In this paper, we introduce a novel problem, i.e., continual learning jailbreak perturbation patterns, into the toxicity detection field. To tackle this problem, we first construct a new dataset generated by 9 types of perturbation patterns, 7 of them are summarized from prior work and 2 of them are developed by us. We then systematically validate the vulnerability of current methods on this new perturbation pattern-aware dataset via both the zero-shot and fine tuned cross-pattern detection. Upon this, we present the domain incremental learning paradigm and the corresponding benchmark to ensure the detector's robustness to dynamically emerging types of perturbed toxic text. Our code and dataset are provided in the appendix and will be publicly available at GitHub, by which we wish to offer new research opportunities for the security-relevant communities.