mirror of https://github.com/hanzoai/hips.git synced 2026-11-03 00:32:50 +00:00

Files

T

zeekay ad5a010f79 Three words for what a proposal is, and a generator that refuses

The same fact lived in nine hand-kept places and they disagreed about 58
proposals. This makes the file the only authority and everything else a
projection of it.

ONE VOCABULARY, in vocabulary.json, read by the lint, the generator and the
doc site. Three words, each answering "can I rely on this?" and nothing else:

    Draft    nothing implements it yet, or its own text says the work is ahead
    Final    the thing it specifies exists in the code
    Living   governance and process: amended in place, never finished

Eight spellings folded in. `Active` was the one doing real damage: on four Meta
and Process HIPs it meant living, on 48 others it meant shipped, and one word
doing two jobs is why nothing agreed. `Review` and `Last Call` described a
queue rather than a document, and a proposal under review is still one you
cannot build against.

STATUS IS EARNED, not declared. Every Standards Track proposal was checked
against the estate: 235 have their implementation, and eight do not and say so.
HIP-0095 carries a twelve-week plan ending at "mainnet launch" and has no code.
HIP-0525 specifies an all-machine meeting while apps/meet refuses a machine a
seat. HIP-0078 names a prover at a "planned location" that does not exist.
HIP-0114 is the cog.* procedure set, and that procedure set returns nothing.
HIP-0003, HIP-0102, HIP-0305 and HIP-0901 likewise: looked, found nothing.

SIXTEEN TOMBSTONES DELETED rather than statused. A dead document kept alive in
the index is one concept wearing two names. Each successor exists and carries
the text, and the 90 references that pointed at a tombstone now point at it --
whole markdown links at once, so link text cannot outlive its target. Three of
the sixteen were pure compatibility: HIP-0028 specified a migration off a
Bitnami chart, HIP-0029 dual-engine MySQL, HIP-0030 a schema-deprecation
policy. We ship no backwards compatibility, so we do not specify it.

Nine links were already broken and had been since the corpus was renumbered:
short-form ./hip-5.md paths that never existed. Two named the wrong proposal --
one cited "HIP-1: Hanzo Multimodal Models" where multimodal is HIP-0003.

THE GENERATOR REFUSES more often than it writes, because a short table looks
exactly as authoritative as a complete one. Rows must equal files; a shrinking
index must say by how much; a section whose input is missing is an error, not
an omission; nothing is written until all of it validates. Its self-test breaks
the corpus eleven ways and requires each to refuse AND leave README byte-
identical -- a guard that reports after half-writing has protected nothing.
The lint's second implementation of the same comparison is gone.

The site rendered 52 proposals with a fallback grey badge and counted them
toward none of its three stat tiles, because docs/lib/source.ts declared a
status union by hand: it admitted a `Stagnant` no file has ever carried and
omitted the one 52 files did. Five components each kept their own copy of the
colours. All six now read vocabulary.json.

2026-08-28 18:05:41 -07:00

12 KiB

Raw Permalink Blame History

hip, title, author, type, status, created, requires

hip	title	author	type	status	created	requires
250	Sustainability Standards Alignment Matrix	Hanzo AI Team	Meta	Living	2025-12-16	HIP-0200

HIP-0250: Sustainability Standards Alignment Matrix

Abstract

This HIP provides a comprehensive mapping between Hanzo AI's Responsible AI (HIP-200) and sustainability frameworks to global standards including NIST AI RMF, ISO/IEC 42001, EU AI Act, environmental standards, and AI-specific governance frameworks. This matrix enables stakeholders to understand how Hanzo's commitments align with regulatory and voluntary standards.

Purpose

AI companies face increasing scrutiny from:

Regulators: EU AI Act, emerging US AI regulations
Enterprise customers: AI governance requirements in procurement
Investors: ESG criteria for AI investments
Partners: Interoperability and trust requirements

This matrix demonstrates our commitment to meeting the highest standards.

AI Governance Standards

NIST AI Risk Management Framework (AI RMF)

NIST Function	Category	Hanzo Implementation	HIP Reference
GOVERN	1.1 Legal compliance	Legal review process	HIP-200
	1.2 Trustworthy AI characteristics	Core principles defined	HIP-200
	1.3 Workforce diversity	Hiring practices	HIP-200
	1.4 Risk culture	Safety-first culture	HIP-200
	1.5 Risk tolerance	Risk appetite defined	HIP-201
MAP	2.1 System context	Use case documentation	HIP-201
	2.2 Impact assessment	Pre-deployment review	HIP-201
	2.3 Trustworthiness characteristics	Model cards	HIP-240
MEASURE	3.1 Risk metrics	Safety metrics dashboard	HIP-210
	3.2 Qualitative analysis	Red team exercises	HIP-210
	3.3 Tracking mechanisms	Production monitoring	HIP-210
MANAGE	4.1 Risk prioritization	Risk matrix	HIP-201
	4.2 Risk treatment	Mitigation strategies	HIP-201
	4.3 Incident response	24/7 response capability	HIP-200

ISO/IEC 42001 (AI Management System)

ISO Clause	Requirement	Hanzo Implementation	Status
4. Context	4.1 Understanding organization	Mission & strategy documented	Implemented
	4.2 Interested parties	Stakeholder mapping	Implemented
	4.3 Scope	AI system boundaries defined	Implemented
	4.4 AI MS	Management system established	In Progress
5. Leadership	5.1 Leadership commitment	Board oversight	Implemented
	5.2 AI policy	HIP-200 serves as policy	Implemented
	5.3 Roles & responsibilities	RACI matrix defined	Implemented
6. Planning	6.1 Risk/opportunity	Risk register maintained	Implemented
	6.2 AI objectives	Safety metrics defined	Implemented
7. Support	7.1 Resources	Dedicated safety team	Implemented
	7.2 Competence	Training requirements	In Progress
	7.3 Awareness	All-hands AI ethics training	Planned
	7.4 Communication	Transparency reports	Implemented
8. Operation	8.1 Planning & control	Model release process	Implemented
	8.2 AI risk assessment	Pre-deployment review	Implemented
	8.3 AI risk treatment	Mitigation implemented	Implemented
	8.4 AI system lifecycle	Full lifecycle governance	In Progress
9. Evaluation	9.1 Monitoring	Continuous monitoring	Implemented
	9.2 Internal audit	Quarterly internal review	Implemented
	9.3 Management review	Board-level review	Implemented
10. Improvement	10.1 Nonconformity	Incident response process	Implemented
	10.2 Continual improvement	Iterative enhancement	Implemented

Certification Status: Target Q4 2025

EU AI Act Compliance

Risk Category	Requirements	Hanzo Compliance	Status
Prohibited	No prohibited use cases	Policy excludes prohibited uses	✅ Compliant
High-Risk	Full compliance suite required	Full controls where applicable	In Progress
	Risk management system	HIP-201 MRM	✅
	Data governance	HIP-205 Data Governance	✅
	Technical documentation	Model cards (HIP-240)	✅
	Record keeping	Audit trails	✅
	Transparency	User disclosure	✅
	Human oversight	HIP-230	✅
	Accuracy & robustness	HIP-210 Safety Evaluation	✅
Limited Risk	Transparency obligations	Disclosure implemented	✅ Compliant
Minimal Risk	Voluntary codes	Adhering to HIP framework	✅ Compliant

OECD AI Principles

OECD Principle	Definition	Hanzo Implementation	HIP Reference
Inclusive Growth	AI benefits broadly shared	Open source models, API access	HIP-200
Human-Centered Values	Respect human rights & democracy	Core principles	HIP-200
Transparency	Meaningful information about AI	Model cards, disclosures	HIP-240
Robustness	Security, safety, oversight	Safety evaluation suite	HIP-210
Accountability	Clear ownership	Named owners per system	HIP-200

Environmental & ESG Standards

GHG Protocol (AI Operations)

Scope	Source	Hanzo Reporting	HIP Reference
Scope 1	Direct emissions	None (cloud-based)	N/A
Scope 2	Purchased electricity	Training & inference compute	HIP-260
Scope 3	Value chain	Hardware manufacturing, data centers	HIP-260

Green AI Metrics

Metric	Definition	Hanzo Tracking	Target
Training emissions	tCO2e per model	Tracked per training run	-30% YoY
Inference efficiency	Tokens per kWh	Real-time monitoring	+50% YoY
Carbon intensity	gCO2e per 1M tokens	Published in model cards	Report
PUE impact	Power Usage Effectiveness	Data center selection criteria	<1.2

AI Environmental Impact Standards

Framework	Element	Hanzo Approach
ML Emissions Calculator	Estimate training emissions	Integrated into training pipeline
Code Carbon	Real-time tracking	Planned integration
Green Algorithms	Energy estimation	Benchmarking tool
IEA Guidelines	Data center efficiency	Vendor selection criteria

Security & Privacy Standards

SOC 2 Type II (AI-Specific)

Trust Principle	AI-Specific Control	Hanzo Implementation	Status
Security	Model access control	Role-based API access	Implemented
	Training data protection	Encrypted storage	Implemented
	Adversarial protection	Input validation	Implemented
Availability	Model serving SLA	99.9% uptime target	Implemented
	Failover capability	Multi-region deployment	Implemented
Confidentiality	Prompt confidentiality	No training on user data	Implemented
	Output confidentiality	Logging controls	Implemented
Privacy	PII handling	HIP-270 compliance	Implemented
	Data minimization	Collection limits	Implemented

Certification Status: Target 2025

ISO 27001 (Information Security)

Domain	Control	Hanzo Implementation
A.5 Information Security Policies	Policy documentation	HIP-200 series
A.6 Organization of InfoSec	Security team structure	Dedicated team
A.8 Asset Management	AI system inventory	Maintained registry
A.9 Access Control	API key management	Implemented
A.12 Operations Security	Monitoring & logging	24/7 SOC
A.14 System Development	Secure SDLC	AI-specific controls
A.16 Incident Management	Incident response	HIP-200 procedures

Certification Status: Target 2025

Privacy Frameworks

Framework	Requirement	Hanzo Compliance
GDPR	Art. 22 (automated decisions)	Human oversight option
	Art. 13-14 (transparency)	Disclosure provided
	Art. 17 (right to erasure)	Data deletion capability
	Art. 35 (DPIA)	Impact assessments
CCPA	Right to know	Disclosure available
	Right to delete	Deletion capability
	Right to opt-out	Opt-out mechanism
ISO 27701	Privacy management	Planning certification

AI-Specific Voluntary Standards

Partnership on AI Tenets

Tenet	Commitment	Hanzo Implementation
Safety-critical AI	Rigorous testing	HIP-210 evaluation suite
Fairness & inclusivity	Bias testing	HIP-220 testing framework
Transparency	Publish research	Open publications
Labor & economy	Responsible deployment	Workforce considerations
Collaboration	Industry engagement	Active PAI member

Anthropic's Responsible Scaling Policy

Commitment	Description	Hanzo Alignment
ASL-1	Basic safety	Baseline for all models
ASL-2	Enhanced safety	Current production standard
ASL-3	Advanced safety	Commitment for future capability
Eval-driven	Evaluation before capability	Pre-deployment gates

Model Cards (Mitchell et al.)

Model Card Section	Required Info	Hanzo Implementation
Model Details	Architecture, training	Included
Intended Use	Use cases, users	Included
Factors	Relevant attributes	Included
Metrics	Evaluation metrics	Included
Evaluation Data	Test datasets	Included
Training Data	Data description	Included
Quantitative Analyses	Performance results	Included
Ethical Considerations	Risks, mitigations	Included
Caveats	Limitations	Included

Standards Compliance Summary

Full Alignment

Standard	Status	Evidence
NIST AI RMF	Aligned	HIP-200 series
OECD AI Principles	Aligned	Policy documentation
Partnership on AI	Member	Public commitment
Model Cards	Implemented	Published with releases

In Progress

Standard	Gap	Timeline
ISO/IEC 42001	Certification process	Q4 2025
SOC 2 Type II	Audit engagement	2025
EU AI Act	Full high-risk compliance	2025

Planned

Standard	Dependency	Target
ISO 27001	Security controls formalization	2025
ISO 27701	Privacy program maturity	2026

Using This Matrix

For Enterprise Customers

Reference HIP numbers in vendor assessments and procurement questionnaires.

For Regulators

This matrix demonstrates proactive compliance with emerging AI regulations.

For Partners

Map integration requirements to Hanzo's documented standards.

For Auditors

Use this as a starting point for compliance verification.

HIP-200: Responsible AI Principles and Commitments (parent document)
HIP-201: Model Risk Management (MRM)
HIP-205: Data Governance & Consent
HIP-210: Safety Evaluation Suite
HIP-220: Bias & Fairness Testing
HIP-230: Human Oversight & Escalation
HIP-240: Transparency Reports
HIP-251: Green AI Compute Practices
HIP-260: Carbon-Aware Training & Inference
HIP-270: Privacy-Preserving ML

Changelog

Version	Date	Changes
1.0	2025-12-16	Initial draft

12 KiB Raw Permalink Blame History