Legacy Fraud Systems Block 96% of Legitimate AI Agent Transactions, Study Finds

Modern AI agents executing automated payments are being flagged as malicious bots by traditional fraud detection systems at a false positive rate of nearly 96%, according to data cited by MRC research. The core problem is that legitimate agent behavior — fast, repetitive, and fully automated — is behaviorally indistinguishable from bot attacks under rules designed for human users. An emerging framework called Know Your Agent (KYA) proposes solving this through cryptographic identity verification, confirming whether an agent is registered, who authorized it, and whether a transaction falls within its approved scope. If all three checks pass, the transaction is treated as legitimate regardless of how automated it appears. The KYA standard is gaining traction in the context of post-MiCA regulated agent commerce, where accountability and delegation provenance are increasingly required.