CertiK at IDAI Summit 2026: AI Adoption & Digital Asset Cybersecurity

**CertiK warns that rapid AI agent adoption in Web3 has produced systemic security failures—researchers found widespread malicious OpenClaw skills and a high-profile agent error that led to an irreversible $450k token transfer—highlighting architectural blind spots (validation/execution divergence, fragile multi-channel identity binding, and persistent memory poisoning). The report describes attack patterns, real incident findings, and recommended institutional controls (least privilege, masking, human-in-the-loop multisig, and AI-driven continuous security), and advertises CertiK's AI Auditor as a mitigation tool.**