Principles

Foundational ideas that anchor the framework and should change rarely.

These principles anchor the HALOS framework. They are intended to change rarely. Implementations, specifications, and proposals should be evaluated against them.

1. Human Primacy

Humans are the originators of meaningful creative direction. Agents assist, amplify, and execute; they do not substitute for human intent. When in tension, human judgment overrides agent output. Systems designed under HALOS assume humans retain final authority.

2. Ideas as Assets

Ideas and creative works have intrinsic value. They deserve ownership, attribution, and provenance. Treating ideas as first-class assets means:

• Originators can claim authorship
• Lineage can be traced
• Value flows back to those who contribute

This applies whether the output is code, prose, art, or design. The form matters less than the recognition that creation has worth.

3. Attribution and Provenance

Contribution should be traceable. For any work produced through human–agent collaboration:

• Human contributions should be identifiable
• Agent involvement should be disclosed
• Provenance (where ideas came from, how they were combined) should be representable

Attribution is not merely ethical—it enables trust, accountability, and the possibility of fair reward.

4. Transparency of AI Involvement

When AI participates in creation, that participation should be visible. Hidden or undisclosed use of AI undermines trust. Transparency does not require exposing model internals; it requires clarity about when and how agents contributed. Users, collaborators, and downstream consumers should be able to discern AI involvement when it matters.

5. Ethical Guardrails

Human–agent collaboration should operate within ethical boundaries. HALOS does not prescribe a single ethics framework, but it assumes:

• Harmful uses are out of scope
• Deception about AI involvement is unacceptable
• Human agency must not be eroded by design

Systems claiming alignment with HALOS should be able to articulate their ethical stance and how it maps to these guardrails.

6. Evolving Standards, Stable Principles

Standards and specifications may change; principles should not. The framework will evolve to address new capabilities and use cases. Proposals can extend, refine, or even retire specific standards. The principles in this document are the anchor—they change only through deliberate, community-wide process and with strong justification.

7. Governance Through Proposal

The framework evolves through open proposal and review. No single entity controls HALOS. Changes follow a documented process: propose, discuss, revise, decide. Governance is lightweight but explicit. See governance.md for details.

8. Innovation with Accountability

Experimentation is encouraged; recklessness is not. HALOS supports innovation in human–agent collaboration. It also expects that innovators can articulate how their work aligns with (or deliberately departs from) the principles. Accountability is not a brake—it is the condition for sustained trust.

---

These principles are stable. Proposals that would materially alter them require exceptional justification and broad consensus.