Backdoor4Good: Benchmarking Beneficial Uses of Backdoors in LLMs

Backdoor mechanisms have traditionally been studied as security threats that compromise the integrity of machine learning models. However, the same mechanism -- the conditional activation of specific behaviors through input triggers -- can also serve as a controllable and auditable interface for trustworthy model behavior. In this work, we present \textbf{Backdoor4Good (B4G)}, a unified benchmark and framework for \textit{beneficial backdoor} applications in large language models (LLMs). Unlike conventional backdoor studies focused on attacks and defenses, B4G repurposes backdoor conditioning for Beneficial Tasks that enhance safety, controllability, and accountability. It formalizes beneficial backdoor learning under a triplet formulation $(T, A, U)$, representing the \emph{Trigger}, \emph{Activation mechanism}, and \emph{Utility function}, and implements a benchmark covering four trust-centric applications. Through extensive experiments across Llama3.1-8B, Gemma-2-9B, Qwen2.5-7B, and Llama2-13B, we show that beneficial backdoors can achieve high controllability, tamper-resistance, and stealthiness while preserving clean-task performance. Our findings demonstrate new insights that backdoors need not be inherently malicious; when properly designed, they can serve as modular, interpretable, and beneficial building blocks for trustworthy AI systems. Our code and datasets are available at https://github.com/bboylyg/BackdoorLLM/B4G.

Key Contributions

Triplet formulation (T, A, U) — Trigger, Activation mechanism, Utility function — that formalizes beneficial backdoor learning for LLMs
B4G-Bench: a standardized benchmark covering four trust-centric tasks (safety enhancement, style personalization, access control, watermarking/identity verification) across four LLMs
Empirical demonstration that beneficial backdoors can achieve high controllability, tamper-resistance, and stealthiness while preserving clean-task performance

🛡️ Threat Analysis

Model Poisoning

The paper's primary subject is backdoor mechanisms — trigger-conditioned hidden behaviors — repurposed for beneficial goals. The entire framework is built on backdoor injection and activation, making ML10 the direct category regardless of whether the intent is malicious or beneficial.