How Agentic AI Will Reshape Payments

IMF Notes, April 24, 2026

Source details

Agentic AI systems can act autonomously on behalf of users: interpret objectives, break them into tasks, and interact with digital services with limited human input.
In payments, agentic AI could shift transactions from human-initiated instructions to agent-mediated decisions.
Rapid experimentation suggests growing relevance, though agentic systems are still emerging.
The note emphasizes that outcomes will depend on institutional design and governance as much as technology.

Intent: agentic AI interprets user objectives and translates goals into payment-related tasks.
Authorization: agentic systems may mediate or automate authorization decisions, creating a tension between probabilistic AI behavior and deterministic authorization rules.
Settlement: agentic activity interacts with settlement processes where control and finality are critical; design must preserve these properties.

Authorization
- Potential shift toward agent-mediated decisions for initiating payments.
- Need to reconcile probabilistic AI outputs with deterministic authorization requirements.
Liquidity
- Agentic behavior could change intraday liquidity patterns as agents optimize timing and routing of payments.
Settlement
- Preserving finality and control when agents act autonomously is a central concern.
Compliance
- Agentic systems introduce new modalities for AML/CFT and compliance functions (referred to as “Agentic compliance”).
Resilience
- Systemic resilience implications arise from widespread agentic adoption and potential correlated behaviors.

Agentic capabilities may add value while preserving control and finality when applied selectively across the intent–authorization–settlement layers.
Use cases span cross-border payment, e-commerce, and compliance automation.

Traceability: agent-mediated actions may complicate tracing the origin and intent behind payments.
Opacity: AI decision processes may be opaque, undermining explainability for authorization and compliance.
Systemic effects: correlated agent behavior could generate systemic risks in payment systems.
Cybersecurity: agentic interfaces expand attack surfaces and introduce new vectors for exploitation.
Legal uncertainty: questions about liability and legal status of agentic actions in payments remain.

Institutional design and governance choices are as important as technical design for shaping outcomes.
Preserve deterministic requirements of payment infrastructures when integrating probabilistic agentic AI—especially for authorization and settlement finality.
Design traceability and explainability mechanisms to support compliance and oversight.
Anticipate systemic and cybersecurity risks from correlated agent behavior and expanded attack surfaces.
Clarify legal frameworks for agentic actions in payments to reduce uncertainty around liability and enforceability.

How Agentic AI Will Reshape Payments; IMF Note No. 2026/004; April 2026
- How Agentic AI Will Reshape Payments; IMF Note No. 2026/004; April 2026 (Markdown version)
- How Agentic AI Will Reshape Payments; IMF Note No. 2026/004; April 2026 (PDF)