Can Small-Scale Data Poisoning Exacerbate Dialect-Linked Biases in Large Language Models?

TL;DR

This paper demonstrates that small-scale data poisoning targeting dialectal styles can significantly amplify sociolinguistic biases and toxicity in large language models, highlighting the need for dialect-aware safeguards.

Contribution

It introduces a novel style-conditioned poisoning method that reveals how dialectal prompts can trigger harmful biases in LLMs, emphasizing the importance of style-aware evaluation and mitigation.

Findings

Poisoned models show increased toxicity for dialectal inputs, especially AAVE.

Conventional toxicity detectors underestimate sociolinguistic harms.

Poisoning can cause emergent jailbreaking behaviors without explicit slurs.

Abstract

Style-conditioned data poisoning is identified as a covert vector for amplifying sociolinguistic bias in large language models. Using small poisoned budgets that pair dialectal prompts -- principally African American Vernacular English (AAVE) and a Southern dialect -- with toxic or stereotyped completions during instruction tuning, this work probes whether linguistic style can act as a latent trigger for harmful behavior. Across multiple model families and scales, poisoned exposure elevates toxicity and stereotype expression for dialectal inputs -- most consistently for AAVE -- while Standard American English remains comparatively lower yet not immune. A multi-metric audit combining classifier-based toxicity with an LLM-as-a-judge reveals stereotype-laden content even when lexical toxicity appears muted, indicating that conventional detectors under-estimate sociolinguistic harms. Additionally, poisoned models exhibit emergent jailbreaking despite the absence of explicit slurs in the poison, suggesting weakened alignment rather than memorization. These findings underscore the need for dialect-aware evaluation, content-level stereotype auditing, and training protocols that explicitly decouple style from toxicity to prevent bias amplification through seemingly minor, style-based contamination.