SybilGuard AI

SybilGuard AI addresses a critical challenge in Web3—Sybil attacks, where malicious actors create multiple wallets to unfairly claim rewards during airdrops or manipulate governance systems. Our platform allows users or protocols to upload a list of wallet addresses, which are then analysed using machine learning algorithms and graph theory to identify connections and suspicious activity patterns.

The AI model analyses historical transaction data to detect clusters of wallets that exhibit Sybil-like behavior, such as high transaction frequency between wallets or shared connections. Flagged addresses are presented with a risk score, enabling protocols to make informed decisions about excluding potential attackers from their airdrop distributions.

By offering a transparent, automated, and scalable solution, SybilGuard AI helps safeguard the integrity of decentralised ecosystems, ensuring fair distribution and preventing manipulation. Our tool can be easily integrated into any protocol running reward-based events, staking, or decentralised governance voting.

Backend & Data Handling: Using python script, the backend fetches transaction data from blockchain APIs (currently using EtherScan). This data is then analysed for suspicious patterns.

AI & Machine Learning: We use Python and graph analysis tools like NetworkX to detect connections between wallet transactions. Machine learning models analyse transaction patterns to identify Sybil attacks.

Graph-Based Detection: Using Graph Theory, we map wallet relationships and look for suspicious structures, such as clusters of wallets interacting abnormally.