New Rules for AI Agent Payments: Large Transactions Must Be Traceable, Pre-Settlement Blocking Becomes Standard

Regulators this week issued compliance guidelines for AI agent payment scenarios, requiring that AI must not initiate payments without explicit user authorization, and that single transactions exceeding a set threshold must be recorded and traceable on a per-transaction basis. The new rules also require payment service providers (PSPs) to conduct pre-settlement screening for transactions initiated by AI agents, including verifying payee whitelists, daily cumulative limits, and sanctions lists.

The guidelines apply to all payment channels integrated with AI agents, covering both stablecoin and fiat settlement. Industry analysts suggest the new rules will drive the adoption of "Know Your Agent" (KYA) standards, requiring payment institutions to verify an agent's identity, authorization intent, and historical behavior, rather than relying solely on traditional KYC. Several leading PSPs have already begun adjusting their systems to incorporate agent payments into existing risk control frameworks.

The new rules do not specify a particular technical solution but reference machine payment protocols such as x402 as a compliance reference. Industry observers note that as AI agents autonomously execute high-frequency, low-value payments like procurement and subscriptions, regulatory focus is shifting from "post-hoc accountability" to "pre-emptive interception," making pre-settlement compliance checks an essential component of payment infrastructure.