Authenticated Workflows: A Systems Approach to Protecting Agentic AI

Mohan Rajagopalan, Vinay Rao · 2026

Read Authenticated Workflows: A Systems Approach to Protecting Agentic AI by Mohan Rajagopalan, Vinay Rao (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-10465.txt.

Description

Mohan Rajagopalan and Vinay Rao propose a trust layer for enterprise AI agents in a 2026 paper. Agentic systems automate workflows, but current defenses like guardrails and meaning filters are chance and routinely bypassed. The paper argues security must instead rest on four edges: prompts, tools, data, and context.

Their authenticated workflows enforce intent and integrity at every crossing, combining cryptographic proof with runtime policy checks. Work either carry valid proof or are rejected, which the authors call deterministic security. A new policy language, MAPL, expresses constraints that evolve as agents and contexts change, scaling far better than flat rule lists. The design aims to make agent security provable rather than hopeful.

Catalog record

CREATORMohan Rajagopalan, Vinay Rao
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE101 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-10465

Browse the full library · About the Tower of Babel