Leveraging Large Language Models and Machine Learning for Smart Contract Vulnerability Detection

S M Mostaq Hossain, Amani Altarawneh and Jesse Roberts · 2025

Read Leveraging Large Language Models and Machine Learning for Smart Contract Vulnerability Detection by S M Mostaq Hossain, Amani Altarawneh and Jesse Roberts (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-02229.txt.

Description

Hossain, Altarawneh, and Roberts wrote this 2025 paper on using AI to find flaws in smart contracts. Smart contracts are self-running agreements on a blockchain, and a bug in one can lock or drain real money. Checking them by hand is slow, so automated detection matters at every step of a deal.

The team trained both classic machine learning models and fine-tuned large language models on a labeled set of contracts. Each contract in the set is marked with its known flaw types. The models learn to sort new code by flaw type. One classic model in the mix is DistilBERT, a compact language model. The authors compare the models head to head. Their bet is that language models tuned just for contract code will catch known flaws, such as reentrancy, better than general tools.

Catalog record

CREATORS M Mostaq Hossain, Amani Altarawneh and Jesse Roberts
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE27 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-02229

Browse the full library · About the Tower of Babel