The Identity-to-Execution Gap
Primary Barrier to Confident AI Agent Adoption in Financial Services
On March 29, 2026, ZLAR founder Vincent Nijjar submitted this paper to the NIST Center for AI Standards and Innovation (CAISI) as part of their Listening Sessions on Sector-Specific Barriers to AI Adoption, focused on financial services.
The submission identifies a single structural gap preventing confident AI agent deployment in regulated financial services: no standard governs what agents do after authentication. The credential boundary is addressed. The execution boundary is not.
The paper presents evidence from the Cloud Security Alliance, the FIFAI II Report, Gartner, and Rubrik Zero Labs quantifying the gap, maps the compounding regulatory friction across SOX, SR 11-7, OCC 2023-17, and BCBS 239, and presents ZLAR as a working solution — an open-source execution-boundary governance system already operating in production.