The HEVIDS Standard for Enterprise AI Governance

Harmony · Ethics · Veracity · Integrity · Discernment · Safeguards

Document identifier	HEVIDS-STD-2026-v1.0
Version	v1.0 — Draft Release, Open for Public Comment
Status	Public draft
Steward	Dr. Anderson D. Prewitt, DRADPA LLC
License (framework text)	Creative Commons Attribution-ShareAlike 4.0 (CC BY-SA 4.0)
Trademark notice	HEVIDS™ is a trademark (U.S. Serial No. 99376104, published for opposition). "AEGIS" and "Sigvis Matrix" are also trademarks. The marks are protected; the standard's text is open.
DOI	https://doi.org/10.5281/zenodo.21045472
Canonical reference text	Prewitt, A. D. (2026). Enhancing Business Value Through Artificial Intelligence. ISBN 979-8-9995924-8-4
Companion formal paper	Prewitt, A. D. (2026). Institutionalizing Trust: HEVIDS and Reverse Game Theory as a Sociotechnical Intervention for Enterprise AI Alignment. SSRN 6915218

---

1. Purpose and Scope

HEVIDS is an open standard for governing the behavior of deployed artificial intelligence systems inside organizations. It addresses a specific and narrow question: once an AI system has been built and approved for use, what governs what it is permitted to do, and how is that governance evidenced?

HEVIDS is deliberately scoped below model training and above runtime engineering. It does not govern how a model is trained — that is the domain of frameworks like Constitutional AI. It does not specify a particular runtime enforcement architecture — that is an implementation choice left to adopting organizations and tooling vendors. HEVIDS specifies the governance outcomes a deployed AI system and its surrounding institutional process must satisfy, expressed as a structured set of functions, categories, and subcategories that an organization can assess against, attest to, and produce evidence for.

The standard is intended for:

• Corporate boards and audit committees exercising AI oversight duties
• Chief AI Officers, Chief Risk Officers, and compliance functions implementing governance programs
• Third-party auditors and assessors evaluating organizational AI governance maturity
• Regulators, standards bodies, and academic researchers studying AI governance frameworks
• Toolmakers building runtime enforcement, monitoring, or audit products that wish to align their outputs to a named external standard

2. Relationship to Other Frameworks

HEVIDS is not a replacement for existing AI risk and compliance frameworks. It is designed to sit alongside them and to be cross-walked against them (see Informative References).

• NIST AI Risk Management Framework — NIST AI RMF defines a four-function risk management process (GOVERN, MAP, MEASURE, MANAGE). HEVIDS defines what a governed AI system's behavior must satisfy; NIST AI RMF defines the organizational process for managing AI risk generally. Organizations following NIST AI RMF can use HEVIDS as the substantive content of their MAP and MEASURE functions for deployed-system behavior.
• ISO/IEC 42001 — ISO 42001 is a certifiable AI management system standard. HEVIDS subcategories are designed to produce evidence artifacts (documented testing, named accountability, retained logs) that map directly onto ISO 42001's clause-level evidence requirements.
• EU AI Act — The EU AI Act imposes binding legal obligations on high-risk AI systems. HEVIDS is not a legal compliance instrument and does not substitute for legal review. It is designed so that an organization implementing HEVIDS in good faith will, as a byproduct, generate much of the documentary evidence the Act's risk-management, record-keeping, and human-oversight articles require.
• Constitutional AI and model-level alignment techniques — These govern what a model is trained to do. HEVIDS governs what a deployed model is permitted to do. The two are complementary; an organization may use both.