authority-hierarchy-is-both-security-mechanism-and-attack-surface

IN derived (depth 2)

Created 2026-06-21T11:28:04+00:00 · Reviewed 2026-06-21T14:41:08+00:00

The LLM instruction authority hierarchy — where system prompts carry higher weight than user messages — is a dual-use security architecture: the same structured privilege boundary that constrains model behavior (defense) creates the privileged channel that prompt injection attempts to impersonate (attack surface), making the security mechanism and the vulnerability architecturally inseparable.

Summary

The same rule that keeps the model in line by giving developer instructions priority over user input is exactly the seam that prompt injection attacks try to fake, so the lock and the hole are the same piece of architecture. This means the threat can't be patched away by fixing a bug; any system built on tiered instruction authority must design its security around that trade-off rather than expecting it to disappear.

Justifications

SL — Authority hierarchy simultaneously constrains behavior and creates injection target

Antecedents (all must be IN):

  • IN system-prompt-higher-authority-than-user — In chat-format LLMs, system prompts carry higher authority than user messages, conveying instructions from a developer or system role.
  • IN prompt-injection-is-architectural-vulnerability-not-bug — Prompt injection is an architectural vulnerability inherent to instruction-following LLMs — arising from the fundamental inability to distinguish developer-defined instructions from user-supplied inputs — rather than a fixable implementation bug, making it the primary and potentially irreducible security concern for deployed applications.

Dependents

These beliefs depend on this one: