PassGPT: Password Modeling and (Guided) Generation with Large Language Models


METADATA ONLY
Loading...

Date

2024

Publication Type

Conference Paper

ETH Bibliography

yes

Citations

Altmetric
METADATA ONLY

Data

Rights / License

Abstract

Large language models (LLMs) successfully model natural language from vast amounts of text without the need for explicit supervision. In this paper, we investigate the efficacy of LLMs in modeling passwords. We present PassGPT, an LLM trained on password leaks for password generation. PassGPT outperforms existing methods based on generative adversarial networks (GAN) by guessing twice as many previously unseen passwords. Furthermore, we introduce the concept of guided password generation, where we leverage PassGPT sampling procedure to generate passwords matching arbitrary constraints, a feat lacking in current GAN-based strategies. Lastly, we conduct an in-depth analysis of the entropy and probability distribution that PassGPT defines over passwords and discuss their use in enhancing existing password strength estimators.

Publication status

published

Book title

Computer Security – ESORICS 2023: 28th European Symposium on Research in Computer Security, The Hague, The Netherlands, September 25–29, 2023, Proceedings, Part IV

Volume

14347

Pages / Article No.

164 - 183

Publisher

Springer

Event

28th European Symposium on Research in Computer Security (ESORICS 2023)

Edition / version

1st Edition

Methods

Software

Geographic location

Date collected

Date created

Subject

Password Guessing; LLMs; Generative AI

Organisational unit

Notes

Funding

Related publications and datasets