16 May 2025

Human Not Present (HNP): The Next Risk Frontier in Payments and Commerce

The transition from traditional 'Card Not Present' (CNP) transactions to a world of autonomous, AI-powered purchasing has opened the door to a new, largely unregulated threat: Human Not Present (HNP) commerce. Unlike CNP, HNP transactions are executed without real-time human input, often driven by digital agents operating on ambient signals or inferred intent. This evolution is redefining risk, liability, and fraud in global commerce. Significant? You bet. Losses from eCommerce CNP fraud is estimated to reach a staggering $28 Billion by next year (2026).

From Card Not Present to Human Not Present

CNP transactions, while risky, still presume a human initiated the purchase. In contrast, HNP transactions delegate purchasing authority to autonomous agents. These systems are trained to act on inferred signals—calendar events, browsing behavior, even prior purchases. Players like Nekuda.ai are pushing this model forward by developing agents capable of purchasing without direct confirmation, based on perceived human intent.

The Role of Payment Infrastructure

Stripe and similar processors are adapting quickly. Stripe’s new modular payment capabilities—multi-party payment flows, programmable checkout, and dynamic refund logic—are designed to support agent-driven commerce. This infrastructure will serve as the backbone of the HNP ecosystem, facilitating real-time, cross-platform, cross-border purchases without direct human intervention.

A Hypothetical Use Case in Travel

Imagine an AI agent notices your favorite band is playing in Barcelona. It checks your calendar, finds a window, and in seconds, books a complete trip: flights from New York, a boutique hotel near the venue, and premium concert tickets. Each purchase is made from a different merchant using different policies and payment processors.

Then the concert is cancelled. Now the problems begin:
- The airfare is non-refundable.
- The hotel charges a no-show fee.
- The ticket is non-transferable.

Who is accountable? What are the consumer’s rights? And how do systems coordinate across merchants to unwind such a transaction?

Risks and Manipulations

With HNP, fraud risk evolves beyond identity theft. New threats include:
- Agent spoofing: Malicious entities pretending to be trusted agents.
- Intent injection: Content manipulation that tricks agents into purchasing unwanted goods.
- Heuristic biasing: Sellers designing offers to exploit decision heuristics in autonomous systems.

These are not hypothetical concerns—they are emerging design patterns.

Recommendations and Regulatory Needs

To manage the risks of HNP commerce, we recommend:
- Establishing agent certification frameworks (analogous to PCI compliance).
- Creating HNP-specific fraud detection models.
- Introducing consumer protection mechanisms like delayed settlement windows.
- Mandating explainability and audit trails for agent-based decisions.
- Facilitating multi-party refund coordination and liability models.

Without regulatory oversight, the promise of agentic commerce may be overshadowed by systemic exploitation and loss of consumer trust.

Conclusion

Agentic AI is revolutionizing commerce. But as we delegate purchasing power to autonomous systems, we must build guardrails to protect consumers, detect abuse, and assign liability clearly. Without proactive intervention, the Human Not Present economy risks becoming a trustless—and dangerous—environment.