AI and the Future of On-Chain Trust and Safety

DeFi security has traditionally focused on post-incident forensics and code-level audits. This talk introduces a new attacker-behavior threat model designed for EVM ecosystems, built on repeatable patterns that appear before an exploit executes. Rodrigo walks through pre-deployment signals, attacker clustering, transaction-intent mismatches, and cross-actor correlations that reveal malicious intent early. The session equips protocols, wallets, and infra teams with practical methods to anticipate and block attacks at the moment of deployment, enabling a more proactive security posture across Ethereum and L2s.