Fraudsters May Target AI Mandates as Agentic Commerce Takes Off

Why It Matters

The shift to mandate‑driven payments creates a high‑value growth channel for banks but also introduces a novel attack surface, making early security adaptation critical for protecting revenue and customer trust.

Key Takeaways

• Agentic commerce could generate $1 trillion US B2C revenue by 2030
• Visa reports 450% rise in dark‑web mentions of “AI Agent”
• Fraud risk shifts to mandate approval and Know‑Your‑Agent verification
• Existing EMV 3‑D Secure, tokenisation can secure agent‑driven payments
• Early adopters can turn mandates into a new always‑on service channel

Pulse Analysis

Agentic commerce, the practice of AI agents executing purchases on behalf of consumers, is poised to reshape retail finance. McKinsey’s forecast of up to $1 trillion in U.S. B2C revenue by 2030 underscores the scale, while global estimates reach $3‑5 trillion. This surge is driven by seamless, always‑on interactions—think a coffee‑monitoring bot that orders beans automatically—eliminating the need for traditional browsers or apps. The market’s rapid expansion is prompting banks to rethink product strategies and invest in the infrastructure that can support such autonomous transactions.

The security implications are equally profound. Visa’s PERC unit flagged a 450% jump in dark‑web chatter about AI agents, prompting a $13 billion allocation toward advanced fraud defenses. Unlike conventional e‑commerce, where risk models rely on device fingerprints and click‑stream data, agentic commerce hinges on cryptographically signed mandates. Fraudsters will therefore aim to hijack the mandate‑creation phase, exploiting weak Know‑Your‑Agent (KYA) checks or social‑engineering tactics to secure overly permissive authorizations. The challenge lies in distinguishing genuine mandate misuse from benign configuration errors, a nuance that will test both issuers and merchants.

Fortunately, the underlying payment stack already contains most of the needed safeguards. EMV 3‑D Secure, tokenisation, and delegated authentication can be extended to validate agent identities and enforce mandate scopes. Banks should upgrade risk engines to ingest agent‑specific protocol data, adopt passkey‑based mandate approvals, and redesign user education around agent registration. Early adopters that integrate these controls will not only mitigate fraud exposure but also unlock a powerful, always‑on service channel that deepens customer engagement and drives new revenue streams.