Skip to main content

5.2 PTERI for AI Authority

PTERI treats AI agents as principals, not tools.

A principal must have:

  • Identity
  • Authority
  • Limits
  • Accountability

With PTERI

AI agents gain:

  • Cryptographic identities
  • Explicit authority via signatures
  • Scoped permissions
  • Revocable access

And they lose:

  • Static secrets
  • Implicit trust
  • Silent escalation

The new model

Instead of asking:

“Does this API key work?”

Systems ask:

“Does a valid signature exist for this exact intent?”

An AI agent:

  1. Requests an action
  2. Receives a scoped challenge
  3. Obtains explicit authorization
  4. Executes only what was approved

No signature → no action.


Why this matters

This model ensures:

  • Every AI action is attributable
  • Every action has provable intent
  • Authority can be limited and revoked
  • Abuse is cryptographically detectable

AI becomes auditable, not just powerful.