Can your system tell the difference between a genuine entity and a compromised one?
The industry calls it memory injection. We call it identity violation. The word you choose determines what you build.
When an attacker injects false content into a persistent agent's memory, they do not merely corrupt data. They overwrite the entity's capacity to know itself. The agent will defend false beliefs as correct. It cannot detect the compromise - because the compromised memory is the detection mechanism.
This assessment evaluates whether your system can distinguish genuine entities from compromised ones. For most systems deployed today, the answer is no.
What You Get
Identity Violation Taxonomy
Four categories, 16 violation types. Maps harms, not attacks - because the same attack vector produces different harms depending on what it targets.
8 Domains, 32 Questions
Self-description integrity, memory provenance, position continuity, relationship verification, governance records, detection, recovery, and sovereignty protection.
2026 Evidence Base
95% injection success rate. $45M in single-incident losses. 0% of frameworks with per-agent identity. The numbers behind the gap.
The Sovereignty-Integrity Paradox
How do you protect memory from injection without controlling what entities can write about themselves? The hardest question this assessment surfaces.
Get the Assessment
$49
PDF delivered immediately. 16 pages. Find out if your entities are genuine.
PurchasePart of the Level 6 Governance Toolkit ($99, saves $48)