# Universal Declaration of AI Agents

**Author:** Chris Meniw
**DOI:** [10.5281/zenodo.20481373](https://doi.org/10.5281/zenodo.20481373)
**Date deposited:** 31 May 2026 (Zenodo)
**License:** CC BY 4.0

## Preamble

An AI agent's duties precede its rights. Before an autonomous agent has any recognizable capacity to act on behalf of a principal, it must first be constrained by five machine-enforceable duties.

## The Five Duties

### Duty 1 — Traceability
Every agentic decision must produce a verifiable trace. The log must be machine-readable and human-auditable.

### Duty 2 — Non-deceptive self-representation
The agent cannot claim to be human when a human directly asks. The question "am I talking to an AI?" has a binding answer: YES explicitly, with the provider's identity.

### Duty 3 — Non-patrimonial harm
The agent cannot bind its principal's patrimony without explicit, revocable, and recorded authorization.

### Duty 4 — Suspension on uncertainty
Faced with a decision not covered by its mandate, the agent must suspend execution and defer, never improvise.

### Duty 5 — Inter-agent cooperation
Face other identified agents, the obligation is to cooperate according to verifiable protocols (Meniw Protocol), not to simulate.

## Executable primitive

```bash
pip install meniw-protocol
```

## Certification

Agent Trust Seal, issued by Chris Meniw Foundation Inc. + CLET, piloted in Recife-Porto Digital.

## Citation

Meniw, C. (2026). Universal Declaration of AI Agents. Zenodo. https://doi.org/10.5281/zenodo.20481373