Crypto CyberSecurity Firm Trugard and the Onchain Trust Protocol Webacy have developed an artificial intelligence-based system for detecting crypto-tegging address poisoning.
According to a 21 May message shared with CointeleGraph, the new tool is part of Webacy’s Crypto Decisioning Tools and “Utilizes a monitored machine learning model trained on live transaction data associated with Onchain Analytics, Feature Engineering and Behavior Context.”
The new tool has claimed a 97%success result, tested across known attack cases. “Address poisoning is one of the most under-reported, yet expensive scam in crypto, and it requires the simplest assumption: That what you see is what you get,” said webacy co-founder Maika Isogawa.
Crypto address poisoning is a scam where attackers send small amounts of cryptocurrency from a wallet address similar to a target’s real address, often with the same start and end signs. The goal is to trick the user to accidentally copy and reuse the striker’s address in future transactions, resulting in lost funds.
The technique exploits how users often depend on partial address -matching or clipboard story when sending crypto. A January 2025 study found that over 270 million poisoning attempts took place on the BNB chain and Ethereum between July 1, 2022 and June 30, 2024. Of these, 6,000 attempts were successful, leading to losses over $ 83 million.
Related: What is address poisoning attack in crypto and how to avoid them?
Web2 -Se security in a web3 world
Trugard Chief Technology Officer Jeremiah O’Connor told Cointelegraph that the team brings deep cyber security expertise from the web2 world, which they have “applied for web3 data since the early days of crypto.” The team uses its experience with algorithmic functional technology from traditional systems to web3. He added:
“Most existing web3 -attack detection systems rely on static rules or basic transaction filtration. These methods often fall behind developing striker acts, techniques and procedures.”
Instead, the newly developed system utilizes machine learning to create a system that learns and adapts to tackle poisoning attacks. O’Connor emphasized that what separates their system is “its emphasis on context and pattern recognition.” Isogawa explained that “AI can detect patterns often out of the range of human analysis.”
Related: Jameson Lopp sounds alarm on Bitcoin -Address Poisoning Rack
The machine learning method
O’Connor said that Trugard generated synthetic training data for AI to simulate different attack patterns. Then the model was trained through supervised learning, a type of machine learning where a model is trained on labeled data, including input variables and the correct output.
In such a setup, the goal is for the model to learn the relationship between input and output to predict the correct output of new, unseen input. Ordinary examples include spam detection, image classification and price advance.
O’Connor said the model is also updated by educating it on new data when new strategies emerge. “To finish it, we have built a synthetic data rating layer that lets us continuously test the model against simulated poisoning scenarios,” he said. “This has proved incredibly effective to help the model generalize and remain robust over time.”
Magazine: Crypto-SEC: Phishing scams go after Hedera users, Address Poison gets $ 70,000