CryptoSentinel: An AI and Blockchain-Based Security Platform for Detecting Crypto Scams and Smart Contract Vulnerabilities
- 1P.E.S. Modern College of Engineering
Abstract
Cryptocurrency and decentralized finance (DeFi) have grown rapidly, but scams, fake tokens, and hacking attacks have caused billions of dollars in financial losses. Because blockchain transactions cannot be canceled or reversed once completed, victims usually have no way to recover stolen money. Existing security tools are either too technical for ordinary users or focus only on one specific problem—such as checking source code or checking web links.
In this project, we designed and developed CryptoSentinel, an easy-to-use and unified cybersecurity platform that protects crypto users from common fraud. CryptoSentinel checks multiple risk factors in real time: (1) wallet addresses across Ethereum, BNB Chain, Polygon, and TRON; (2) smart contracts to find hidden traps like honeypots and excessive taxes; (3) fake peer-to-peer transfers, specifically 'Flash USDT' scams and forged payment receipts; (4) chat messages using a Machine Learning Natural Language Processing (NLP) model to identify seed phrase theft and phishing; and (5) suspicious websites and QR codes using threat intelligence and typosquatting detection. The platform computes a simple Trust Score from 0 to 100 so users can make safe decisions before sending money or connecting their wallets. In our experimental tests on 4,500 messages, 1,200 contracts, and 600 transaction receipts, the system achieved over 97% accuracy in scam message detection and 99% accuracy in finding honeypot contracts, while delivering complete analysis in less than 350 milliseconds.
Related
- Blockchain-enabled IoT security: a multi-layered framework for smart contract vulnerability detection and optimization
- Cybersecurity of Smart Sensing Insole Systems: A Threat Modelling Approach to Understanding Security Vulnerabilities
- BMSVP: Blockchain-based message signing and verification protocol for security enhancement in smart grids
- A Blockchain Platform for Vaccine Beneficiary Registration, Side Effects, Cold Supply Chain, and Personnel Training and Employment Management
- Transient Events: A Rapid-Response, Multi-Platform Observing Strategy for Detecting and Understanding Abrupt Change in Aquatic Ecosystems
BibTeX
@misc{bakale2026cryptosentinel,
title = {{CryptoSentinel: An AI and Blockchain-Based Security Platform for Detecting Crypto Scams and Smart Contract Vulnerabilities}},
author = {Akshay Bakale and Shraddha Nalage and Ritesh Patil},
howpublished = {Zenodo},
year = {2026},
month = oct,
doi = {10.5281/zenodo.23599477},
url = {https://zenodo.org/records/23599477},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - CryptoSentinel: An AI and Blockchain-Based Security Platform for Detecting Crypto Scams and Smart Contract Vulnerabilities
AU - Akshay Bakale
AU - Shraddha Nalage
AU - Ritesh Patil
PY - 2026
DA - 2026/10/23
PB - Zenodo
DO - 10.5281/zenodo.23599477
UR - https://zenodo.org/records/23599477
AB - Cryptocurrency and decentralized finance (DeFi) have grown rapidly, but scams, fake tokens, and hacking attacks have caused billions of dollars in financial losses. Because blockchain transactions cannot be canceled or reversed once completed, victims usually have no way to recover stolen money. Existing security tools are either too technical for ordinary users or focus only on one specific problem—such as checking source code or checking web links. In this project, we designed and developed CryptoSentinel, an easy-to-use and unified cybersecurity platform that protects crypto users from common fraud. CryptoSentinel checks multiple risk factors in real time: (1) wallet addresses across Ethereum, BNB Chain, Polygon, and TRON; (2) smart contracts to find hidden traps like honeypots and excessive taxes; (3) fake peer-to-peer transfers, specifically 'Flash USDT' scams and forged payment receipts; (4) chat messages using a Machine Learning Natural Language Processing (NLP) model to identify seed phrase theft and phishing; and (5) suspicious websites and QR codes using threat intelligence and typosquatting detection. The platform computes a simple Trust Score from 0 to 100 so users can make safe decisions before sending money or connecting their wallets. In our experimental tests on 4,500 messages, 1,200 contracts, and 600 transaction receipts, the system achieved over 97% accuracy in scam message detection and 99% accuracy in finding honeypot contracts, while delivering complete analysis in less than 350 milliseconds.
ER -