Analysis

What Is Prompt Leak? Definition & Examples

Page published

Publication date unknown · First observed: 15 Nov 2026

Coverage timeline

15 Nov 2026nhimg.orgobserved

Single-source analysis — one report is available.

Why it matters

Prompt leakage can expose secrets, internal policies, and tool-access details that help attackers probe AI systems or misuse connected capabilities.

NHI Mgmt Group’s glossary defines prompt leakage as unintended exposure of hidden AI instructions, context, or sensitive data through user interactions or tool responses. The entry describes common exposure paths and general mitigations for agentic and MCP-connected workflows, without reporting a specific incident or new experimental findings.