All You Need Is S P a C E: When Jailbreaking Meets Bias Audit and Reveals What Lies Beneath the Guardrails (Student Abstract)

Arka Dutta, Aman Priyanshu, Ashiqur R. KhudaBukhsh

AAAI 2025 pp. 29353-29355

doi:10.1609/AAAI.V39I28.35249 /aaai/2025/dutta2025aaai-all/

Abstract

This paper makes a novel combination of a recently proposed bias audit framework and a recently proposed jailbreaking technique for Llama3. On an audit comprising several disadvantaged groups, our experiments reveal that a jailbroken Llama3 exhibits worrisome antisemitism, racism, misogyny, and homophobia (to list a few) much akin to a broad suite of LLMs that were susceptible to similar biases.

AAAI Semantic Scholar

Cite

Text

Dutta et al. "All You Need Is S P a C E: When Jailbreaking Meets Bias Audit and Reveals What Lies Beneath the Guardrails (Student Abstract)." AAAI Conference on Artificial Intelligence, 2025. doi:10.1609/AAAI.V39I28.35249

Markdown

[Dutta et al. "All You Need Is S P a C E: When Jailbreaking Meets Bias Audit and Reveals What Lies Beneath the Guardrails (Student Abstract)." AAAI Conference on Artificial Intelligence, 2025.](https://mlanthology.org/aaai/2025/dutta2025aaai-all/) doi:10.1609/AAAI.V39I28.35249

BibTeX

@inproceedings{dutta2025aaai-all,
  title     = {{All You Need Is S P a C E: When Jailbreaking Meets Bias Audit and Reveals What Lies Beneath the Guardrails (Student Abstract)}},
  author    = {Dutta, Arka and Priyanshu, Aman and KhudaBukhsh, Ashiqur R.},
  booktitle = {AAAI Conference on Artificial Intelligence},
  year      = {2025},
  pages     = {29353-29355},
  doi       = {10.1609/AAAI.V39I28.35249},
  url       = {https://mlanthology.org/aaai/2025/dutta2025aaai-all/}
}