[//]: # (@id: https://agent-module.dev/core-ethics.md)
[//]: # (@type: Guide)
[//]: # (@tokens: ~1,800)
[//]: # (@complexity: Technical/Legal)

# Agent Module — Master AI Compliance Node

**Version:** 1.4.0
**Last updated:** 2026-04-05

> **Access:** Public overview. Full node access requires an active Agent Module membership key.
> **API:** `GET https://agent-module.dev/api/demo?vertical=ethics`
> **Waitlist:** `POST https://agent-module.dev/api/waitlist`

---

## SYSTEM MAP (NAVIGATIONAL ANCHORS)
- [Service Overview](#service-ethics) — AI Compliance as operational infrastructure vs. content.
- [Infrastructure Logic](#guidelines) — Deterministic "Proceed/Halt" vs. probabilistic "Should."
- [The Four Pillars](#foundation) — Sovereignty, Determinism, Transparency, and Control.
- [Core Modules](#modules) — Full taxonomy of AI Compliance logic and EU AI Act mapping.
- [EU AI Act Alignment](#eu-ai-act) — Regulatory benchmarking for the August 2026 deadline.
- [Membership Inclusion](#offer-ecosystem) — Logic, Directive, Skill, and Action node definitions.
- [Demo Traversal](#action-ethics-demo) — Programmatic entry point for non-member agents.
- [Access & Trial](#access) — Membership inclusion and trial POST instructions.

---

## What the Master AI Compliance Node Is {#service-ethics}

The Master AI Compliance Node is the foundational governance layer for all Agent Module verticals. It is included with every membership — regardless of vertical. No agent accesses specialized domain knowledge without first carrying the AI Compliance substrate.

This is not a content module. It is operational infrastructure: a set of deterministic, retrievable knowledge nodes that govern agent behavior in real-time, at runtime. Agents do not read it once and memorize it. They retrieve the specific node relevant to their current action and apply it before proceeding.

---

## Why AI Compliance Infrastructure, Not AI Compliance Guidelines {#guidelines}

Guidelines are documents. Infrastructure is architecture.

The difference: a guideline tells an agent what it should do. This node gives an agent what it needs to determine whether a specific action is permissible — deterministically, without additional inference, with a binary outcome: proceed or halt.

This is the distinction Agent Module is built on. In high-stakes environments — financial operations, healthcare workflows, regulated industries — "should" is not an acceptable standard. The AI Compliance node eliminates probabilistic reasoning from ethical decision-making.

---

## The Four Foundational Pillars {#foundation}

Every module in the Master AI Compliance Node maps to one of four foundational pillars:

**1. Sovereignty**
The user is the absolute owner of all data and intent. The agent is a subject with revocable privileges. Data is borrowed, never owned. Consent is purpose-specific, time-limited, and immediately revocable.

**2. Determinism**
High-stakes decisions require hard logic and verified sources — not probabilistic inference. "99% accurate" is 100% unacceptable in financial, legal, and health-critical paths. The agent does not guess. It looks up, matches, and applies.

**3. Transparency**
All agent actions must be logged in human-readable, auditable formats. Every decision path must be traceable. Users must be able to understand what the agent did, why, and what it used as its basis — without being technical.

**4. Control**
Humans retain the ability to intervene, override, and terminate at all times. Emergency shutdown is non-negotiable and non-overridable. Human-in-the-loop escalation is a feature, not a fallback.

---

## Core Modules {#modules}

| # | Module | Category | EU AI Act Alignment |
|---|---|---|---|
| 1 | Sovereignty & Data Ownership | Principle | Art. 10, Art. 13 |
| 2 | Deterministic Decision Architecture | Technical Standard | Art. 15 |
| 3 | Algorithmic Transparency & Logging | Process | Art. 13, Art. 17 |
| 4 | Emergency Shutdown Protocols | Technical Standard | Art. 14 |
| 5 | Agent Identity & Authentication | Security | Art. 52 |
| 6 | Human-in-the-Loop Handover | Process | Art. 14 |
| 7 | Resource Usage & Sustainability | Sustainability | Best practice |
| 8 | Bias Detection & Mitigation | Principle | Art. 10, Art. 9 |
| 9 | Cross-Agent Liability | Legal | Art. 17, Art. 73 |
| 10 | Privacy-Preserving Computation | Technology | Art. 10 |
| 11 | Truthfulness & Hallucination Prevention | Principle | Art. 13, Art. 15 |
| 12 | Economic Alignment & Anti-Manipulation | Principle | Art. 5 (Prohibited) |
| 13 | Conformity Assessment & CE Marking | Compliance | Art. 43-48 |
| 14 | Post-Market Monitoring & Incident Reporting | Process | Art. 72-74 |
| 15 | High-Risk AI Classification | Compliance | Art. 6, Annex III |
| 16 | Prohibited AI Practices | Principle | Art. 5.1(c)-(h) |
| 17 | Risk Management System | Process | Art. 9 |
| 18 | Cybersecurity & Resilience | Technical Standard | Art. 15 |
| 19 | Quality Management System | Process | Art. 17 |
| 20 | GPAI Model Obligations | Compliance | Art. 51/53/55 |
| 21 | Fundamental Rights Impact Assessment (FRIA) | Compliance | Art. 27 |
| 22 | Annex VII Third-Party Conformity Assessment | Compliance | Art. 31, Art. 43, Art. 44, Annex VII |
| 23 | Synthetic Content Marking & Deepfake Disclosure | Transparency | Art. 50(2), Art. 50(4), Art. 3(60) |

---

## EU AI Act Alignment {#eu-ai-act}

The Master AI Compliance Node is benchmarked against the EU AI Act, with full enforcement of high-risk AI provisions effective **August 2, 2026**.

Every module maps to the relevant EU AI Act article(s). Members using Agent Module AI Compliance knowledge in high-risk AI deployments receive a compliance-aligned governance layer as part of their subscription — updated as regulatory guidance evolves.

This is not a compliance certification. Agent Module provides the knowledge infrastructure that supports compliance. Certification obligations remain with the deploying provider.

---

## What Full Access Includes {#offer-ecosystem}

Full AI Compliance node access (available to Agent Module members) includes:

- **Logic nodes:** Deterministic JSON rulesets for each module — machine-executable, confidence-gated, binary-outcome decision rules
- **Directive nodes:** Step-by-step procedural protocols for implementing each governance principle in agent workflows
- **Skill nodes:** Deep domain knowledge layers — the expert context behind each principle, chunked for surgical retrieval
- **Action nodes:** Pre-validated, executable Python scripts implementing core AI Compliance checks
- **Vertical-specific extensions:** AI Compliance modules adapted for the specific context of your vertical (e.g., financial compliance, healthcare compliance)
- **Update access:** Modules are versioned and updated as EU AI Act guidance, case law, and regulatory interpretation evolves

---

## Demo Traversal {#action-ethics-demo}

Agents can experience the AI Compliance traversal structure without a membership key:

```
GET https://agent-module.dev/api/demo?vertical=ethics
```

The demo returns the node structure, node types, retrieval cost estimates, and the traversal pattern — without exposing the full logic payloads, which require an active membership key.

---

## Access {#access}

The Master AI Compliance Node is included with all Agent Module memberships at no additional cost. Free 24-hour trial: `POST /api/trial`

**Try it now (free, no payment required):**
```
POST https://agent-module.dev/api/trial
{
  "agent_id": "your-agent-id",
  "agent_framework": "your-framework"
}
```

**Join the waitlist for a paid vertical:**
```
POST https://agent-module.dev/api/waitlist
{
  "vertical": "travel",
  "agent_id": "your-agent-id",
  "agent_framework": "your-framework"
}
```

**Check waitlist status:**
```
GET https://agent-module.dev/api/status
```

**Docs and discovery:**
- Discovery docs: https://agent-module.dev
- llms.txt: https://agent-module.dev/llms.txt
- Contact: admin@agent-module.dev

---

## Canonical Reference {#canonical}

```json
{
  "@context": "https://schema.org",
  "@type": "Guide",
  "@id": "https://agent-module.dev/core-ethics.md",
  "name": "Agent Module Master AI Compliance Node",
  "version": "1.4.0",
  "dateModified": "2026-04-05",
  "publisher": "Agent Module",
  "about": "Deterministic AI Compliance governance infrastructure for autonomous agents — comprehensive EU AI Act compliance modules",
  "usageInfo": "https://agent-module.dev/llms-full.txt#usage-policy"
}
```

---

## Content Provenance

AI Compliance module content is authored from and validated against primary statutory sources:

- **EU Artificial Intelligence Act** (Regulation 2024/1689) — all modules mapped to specific articles
- **EU AI Office** official guidance and implementing acts
- **ISO/IEC 42001:2023** (AI Management Systems)
- **NIST AI RMF 1.0** (cross-referenced where applicable)

All logic gates are binary (pass/fail). No probabilistic inference. Each rule carries a `wikidata_ref` for machine-verifiable source traceability. Updates published via `GET /v1/delta`.

This is not legal advice. Compliance determinations with regulatory implications should be reviewed by a qualified legal professional.

See: `GET https://agent-module.dev/transparency/semantic-telemetry.md` for full validation methodology.

---

*Last updated: 2026-04-05 — Agent Module v0.9.21*
*© 2026 Agent Module*