MeitY moves to mandate human-in-the-loop controls for agentic AI payments on UPI

CERT-In/MeitY proposes mandatory human interventions and audit trails for AI-agent payments above financial thresholds, as NPCI builds a Unified Agent Protocol for AI-driven UPI transactions—setting guardrails for emerging agentic commerce across Indian retail rails.

— Source publishedThu, 16 Jul, 2026, 13:52 IST·First seen Thu, 16 Jul, 2026, 14:02 IST·Source Medianama

What happened

UPI · CERT-In/MeitY proposes mandating human-in-the-loop controls with audit trails for agentic AI payments above financial thresholds, as NPCI builds a Unified

Why this matters

NPCI's Unified Agent Protocol and MeitY guardrails could reshape the agentic-payments stack, making compliance-ready checkout and audit-infrastructure players attractive tuck-in targets ahead of the mandate.

What to watch

  • MeitY/CERT-In publication of specific financial thresholds and audit retention rules
  • NPCI release of Unified Agent Protocol spec or sandbox availability
  • First PSP or major retailer announcing UAP integration
  • Public consultation deadlines and industry body (NASSCOM/MPF) response
  • Any card network or wallet adopting parallel agentic controls
  • Map current agentic checkout flows against likely threshold tiers and flag transactions needing step-up confirmation
  • Engage NPCI/PSP partners on Unified Agent Protocol early-access and certification timelines
  • Build modular consent + audit-trail logging into agent orchestration layer now to avoid retrofit
  • Draft dual-lane UX (auto vs. human-confirm) prototypes to preserve conversion at high-value carts
  • Assign compliance owner to track CERT-In/MeitY draft consultation windows and submit feedback