Core Concepts in Authority Architecture

An authority surface describes the complete set of consequential actions a system can produce — whether by intent, configuration, or emergent behavior. It differs from attack surface in that it encompasses all functional capability, not just exploitable vulnerabilities.

A system with a large authority surface is dangerous even if it is entirely uncompromised, because its legitimate capabilities represent a potential amplification vector for any downstream attacker who gains influence over it.

The synthetic insider is a new threat model category. Unlike a compromised employee or a malicious external attacker, a synthetic insider is an AI agent or automation pipeline that holds legitimate authority and processes untrusted input.

It cannot be stopped by identity checks alone because it is already authorized. The risk is not credential theft — it is behavioral manipulation. Prompt injection, data poisoning, and adversarial input all represent attacks on synthetic insider judgment rather than on access controls.

Authority amplification occurs when the compromise or manipulation of one system produces a disproportionate expansion of available authority. This is the core mechanism by which modern AI-integrated attacks cause outsized damage.

A low-privilege customer service chatbot that can initiate helpdesk tickets that trigger privileged backend automation has a dramatically amplified authority surface. Understanding amplification requires mapping the full graph of authority transit between systems, not just the permissions of each system in isolation.