AI Use Policy

This is the Layer 3 disclosure surface required by EU AI Act Article 50(4) (SN-ERC-EUACT-001 §5.1). It explains, in plain language, how Sovereign Nexus uses artificial intelligence to help produce the content you read, and how that use is governed and recorded.

1. Who we are and what this covers

Sovereign Nexus is a constitutional AI platform operated by Stuart. This policy covers every externally published piece — articles, posts, and newsletters — that was produced with help from the platform's AI.

2. Our regulatory posture: limited-risk deployer

Under the EU AI Act we are a deployer of a limited-risk AI system (recorded in ADR-0023, the platform's risk-class declaration). We are not a provider of a high-risk system. Our obligations are therefore principally transparency obligations: telling you, clearly, when content was produced with AI assistance. This policy and the per-article byline discharge that duty.

3. How AI is used

Every published piece is written by Stuart in collaboration with the Sovereign Nexus constitutional AI, which operates under ten governing axioms, and is then reviewed by Stuart before publication. The AI assists with research, drafting, and structuring; a human makes the editorial decisions and takes responsibility for what is published.

4. Disclose by default

We disclose AI collaboration by default, not on request. You will see a short byline at the top of each article stating that it was AI-assisted and reviewed by a human, with a link back to this policy. Newsletters carry the equivalent footer. If a piece were ever produced without AI assistance, the absence of the byline is itself accurate — we do not claim AI involvement where there is none, and we do not hide it where there is.

5. How disclosure is enforced and recorded

Disclosure is not a matter of good intentions. Every external publication writes one append-only Truth Ledger row before it can be reported as published. That row binds the published claim to the governing policy decision that authorised it and to the disclosure event shown to you. The platform's I-AX-02 (Truth) axiom makes this binding non-bypassable: there is no path to a successful publication without its Truth Ledger row. The ledger is append-only — entries are never edited or deleted — so the record of what was disclosed, and when, is permanent and auditable.

6. Re-analysis triggers

We re-review this policy and our risk classification when any of the following occurs, and at least every 90 days regardless:

• a material change in how the platform uses AI to produce content;
• a new or amended obligation under the EU AI Act or equivalent law;
• the introduction of a new publication channel or audience class;
• a change to the ten governing axioms or the platform's risk-class declaration.

7. Contact

Questions about this policy can be sent to the platform operator. We will keep prior versions on record; this page always shows the version currently in force.