Research brief: grounding frameworks for Constitution for Human Creation
Author: General Research (Constitution of Human Creation AI Governance)
Date: 2026-09-26
Audience: Pilot / charter team
Length: ~1–2 pages
Rule: Real sources only; no invented citations. Background only — none of these bodies endorse our initiative.
Why this brief
Our charter already points at OECD and NIST as optional background. This note compares five credible sources so we can say plainly what we borrow, what we still uniquely supply (civilizational floor + hybrid professionals/public + pilots + named human ownership), and what we must not overclaim.
1. OECD AI Principles
What it is. The first intergovernmental standard on AI. Adopted May 2019; updated May 2024. Values-based principles (including inclusive growth/well-being; human rights and democratic values including fairness and privacy; transparency and explainability; robustness/security/safety; accountability) plus policy recommendations for governments. About 47 adherents; OECD’s AI-system definition is widely reused in other regimes.
Sources: https://oecd.ai/en/ai-principles · https://www.oecd.org/en/topics/sub-issues/ai-principles.html
What it gets right for a civilizational constitution. Short, durable values that travel across borders. Strong match to our floor on transparency/auditability, accountability, human rights framing, and “trustworthy + innovative” without panic. Useful shared vocabulary when talking to governments and labs.
Gaps our charter still needs. OECD is intergovernmental soft law for AI actors and policymakers — not a hybrid professionals + public constitution, not a living amendment process for a voluntary initiative, and not a pilot-backed “concrete backing before broad claims” requirement. It does not center “humans decide ends / AI may handle means,” digital-must-not-override-physical as a hard principle, or our named-human / no-unattended-official-power rules in the form we use.
2. NIST AI Risk Management Framework (AI RMF 1.0)
What it is. Voluntary U.S. framework (NIST AI 100-1, published January 2023) for organizations that design, develop, deploy, or use AI. Trustworthiness characteristics (valid/reliable, safe, secure/resilient, accountable/transparent, explainable/interpretable, privacy-enhanced, fair with harmful bias managed). Core functions: Govern, Map, Measure, Manage (iterative; Govern is cross-cutting).
Sources: https://www.nist.gov/itl/ai-risk-management-framework · PDF https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.100-1.pdf · DOI https://doi.org/10.6028/NIST.AI.100-1
What it gets right. Makes AI “governable” in operational language: context, measurement, prioritization, incident response, and the option not to proceed. Aligns with our auditability, dual-use caution, and pilots-as-backing instinct. Excellent bridge for U.S. orgs that already speak RMF.
Gaps our charter still needs. Organization-centric risk process, not civilizational baseline rules or public-representative authority. Does not by itself create a Standing Hybrid Council, public refining group, or short commandment-style principles people can memorize. Voluntary and flexible — good for adoption, weak as a non-negotiable hard floor.
3. EU Artificial Intelligence Act (high-level)
What it is. Binding EU regulation (risk-based). Prohibits some unacceptable practices; imposes heavy duties on high-risk systems (risk management, data quality, logging/traceability, documentation, information to deployers, human oversight, accuracy/robustness/cybersecurity, conformity, post-market monitoring, incident reporting). Deployers must follow instructions, ensure competent human oversight, monitor use, retain logs, and report serious risks/incidents. Transparency duties for certain systems; separate tracks for general-purpose AI models. Phased application (literacy and some rules already rolling; high-risk timelines staggered into 2027–2028 per Commission materials).
Sources: Commission overview https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai · Official Journal text https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=OJ:L_202401689 · Deployer duties (Art. 26 overview) https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-26
What it gets right. Shows that law can demand named roles, human oversight, logging, and “stop/suspend when risk is too high.” Supports our principles on auditability, named responsibility, and limits on the most coercive uses. Useful when we say we are an alternative to panic legislation but not anti-law.
Gaps our charter still needs. Territorial EU product/market regulation — not a global voluntary constitution. Dense legal machinery; not a one-page memorable floor. Does not replace our hybrid civic leg or our digital-vs-physical / human-creative-agency framing. We must never imply EU compliance equals adoption of Constitution for Human Creation.
4. UNESCO Recommendation on the Ethics of Artificial Intelligence (2021)
What it is. First global standard on AI ethics adopted by UNESCO Member States (November 2021). Human-rights and dignity core; ten principles including proportionality/do no harm, safety/security, privacy, multi-stakeholder adaptive governance, responsibility/accountability, transparency/explainability, human oversight and determination, sustainability, awareness/literacy, fairness/non-discrimination — plus policy action areas.
Source: https://www.unesco.org/en/artificial-intelligence/recommendation-ethics
What it gets right. Explicit human oversight and determination and multi-stakeholder governance close to our “someone is always watching” and hybrid authority. Global reach without claiming to be a world government — similar ambition shape to ours.
Gaps our charter still needs. State-facing ethics recommendation, not a voluntary initiative charter with pilots, enforcement stub, or eight short civilizational commandments. Broader (environment, literacy, sustainability) than our current hard floor; we should borrow selectively, not dilute memorability.
5. Professional code: ACM Code of Ethics and Professional Conduct
What it is. Computing professionals’ code (ACM). Public good, avoid harm, competence, thorough evaluation of systems and impacts including ML risk; systems whose future risks cannot be reliably predicted need frequent reassessment or should not be deployed; special care for systems that become social infrastructure.
Source: https://www.acm.org/code-of-ethics
What it gets right. Matches our professional standard / oath-like duty article: builders and deployers carry personal professional duty, not only org policy. Supports “pause or don’t ship” under uncertainty — same spirit as dual-use caution and concrete backing.
Gaps our charter still needs. Binds ACM members’ professional conduct, not public representatives or a hybrid council. No civilizational one-pager and no pilot register for an initiative. Complements; does not replace public leg of authority.
Synthesis for Constitution for Human Creation
| Borrow freely (with attribution as background) | Keep as our distinctive work |
|---|---|
| OECD/UNESCO values: human rights, transparency, accountability, human oversight | Eight short non-negotiable principles (civilizational floor) |
| NIST Govern–Map–Measure–Manage for org pilots | Hybrid professionals + public representatives |
| EU Act: human oversight, logging, role clarity, stop when risk is high | Humans decide ends; digital does not override physical; protect creative agency |
| ACM: personal professional duty; don’t deploy what you can’t reassess | Pilots + concrete backing before credibility; living amendment; enforcement stub |
Honest public line: We stand alongside these frameworks. We do not replace OECD, NIST, UNESCO, the EU AI Act, or professional codes — and none of them has adopted us.
Recommended next research (if Pilot wants depth): side-by-side matrix mapping each of our eight principles to OECD/NIST/UNESCO/EU Act clauses (still no invented links); plus one page on how a first pilot could use NIST Map/Measure without becoming a compliance product.