AI Security Agent Detects 92% of DeFi Smart Contract Vulnerabilities

AI Security Agent Detects 92% of DeFi Smart Contract Vulnerabilities

First seen 21 Feb 2026, 23:11 UTC CoindeskTokenpostWhalesbookSecurityboulevard 30.6

Article Content

Browse articles
ThreatCluster

A specialized AI security agent has successfully detected 92% of vulnerabilities in exploited DeFi smart contracts, according to a benchmark study by AI security firm Cecuro. The analysis focused on 90 real-world smart contracts exploited between October 2024 and early 2026, resulting in verified losses of $228 million. This research indicates a significant advancement in the detection capabilities of purpose-built AI models over general-purpose ones.

Timeline

2024-10-01
Exploits of DeFi smart contracts began
2026-02-20
Coindesk article published on AI detection capabilities
2026-02-21
Whalesbook and Tokenpost articles published on AI benchmark results