AttackPathGNN: Cross-function vulnerability detection in smart contracts using state interference graphs and conjunction pooling
Researchers have developed AttackPathGNN, a novel graph neural network designed to detect vulnerabilities in smart contracts. Unlike previous methods that focus on individual functions, AttackPathGNN analyzes relationships between functions and the conditions that enable exploits. The system utilizes a State Interference Graph to link functions sharing mutable storage and employs conjunction pooling to aggregate exploit preconditions, achieving high accuracy on benchmark datasets. AI
IMPACT Enhances security for smart contracts by providing a more robust detection method for complex, multi-function exploits.