Abstract
Settlement-Time Execution Finality for Agentic, API and Automated Payments
ABSTRACT :
Payment rails can move money, but they cannot tell whether a generated instruction is the one that was authorised. A signed ISO 20022 message, an OAuth token, a mandate, a 3-D Secure result or an idempotency key can each be valid while the act is wrong. Fraud has moved from stealing the instrument to stealing the session that can pay, and now to AI agents that construct payments: an injected invoice can make an authorised accounts-payable or checkout agent pay a new beneficiary or a changed amount. This disclosure holds every payment instruction as a non-effective candidate act. A protected enforcement domain binds payer, amount, currency, beneficiary, rail, purpose, policy epochs and the settlement point into an instruction digest, checks mandates, risk lists, limits and maker-checker approvals, commits evidence, and issues single-use, non-bearer authority. The component that posts or captures funds verifies the live instruction against that authority and consumes it atomically with the posting; any mismatch, replay, stale state or timeout leaves balances unchanged. Coverage includes PSP checkout, core banking, ACH, SEPA Instant, card capture, treasury agents, payroll, FX legs, CBDC and MCP payment tools, with a 52-test reference implementation. Industry-alignment material is AI-generated; readers must cross-check it against the original patent filings before drawing conclusions. This disclosure is based on PCT/IB2026/055615 (the "Mothership" application), filed 4 June 2026 and publicly disclosed 16 July 2026, which claims priority from PCT/IB2026/053385, filed 7 April 2026 and disclosed 4 June 2026. The author is open to acquihire or strategic collaboration. IETF Datatracker: https://datatracker.ietf.org/doc/draft-das-payment-execution-finality/
KEYWORDS payment instruction authorization, settlement-time verification, instruction digest binding, beneficiary substitution prevention, amount substitution prevention, single-use payment authority, consume-once posting, replay protection, business email compromise mitigation, authorised push payment fraud, maker-checker threshold approval, ISO 20022 release gate, agentic payment agent, sender-constrained capability, fail-closed settlement
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License
Recommended Citation
Das, Sangam Kumar, "The AI Agent Was Allowed to Pay. It Paid the Wrong Person: Why Payment Systems Need a Final Check Before Money Moves", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/12804