Home / Standards
AIGS series

Standards catalog

The AIGS series sets out what accountable enterprise AI looks like in practice. Each standard states requirements, the evidence that proves them, and the metrics an organization reports.

Founding drafts are open for comment. All six standards and the Framework are exposure drafts. Nothing here is final until the Standards Board votes after the comment period. How standards are made.
NumberTitleDocumentStatusWorking group
AIGFAccountability Framework
The lifecycle, control objectives and five-level maturity model that tie the series together.
ED-2026-FExposure DraftStandards Board
AIGS 100Accountable Ownership
Every AI system has a named, current owner of record.
ED-2026-100Exposure DraftTWG-100 Ownership
AIGS 200Cost and Token Accountability
AI consumption is attributed to an owner, a purpose and a budget.
ED-2026-200Exposure DraftTWG-200 Cost
AIGS 300Usage and Access
Who uses which AI, and whether it is sanctioned, with stated coverage.
ED-2026-300Exposure DraftTWG-300 Usage
AIGS 400Data Protection and Personal Data
Data reaches AI only where classified, permitted and logged.
ED-2026-400Exposure DraftTWG-400 Data
AIGS 500Agent and Tool Security
Least privilege for agents; exposure paths found and treated.
ED-2026-500Exposure DraftTWG-500 Agents and Tools
AIGS 600Evidence and Assurance
Every claim labeled by evidence strength; no false greens.
ED-2026-600Exposure DraftTWG-600 Evidence
How to read a standard

Every AIGS standard has the same anatomy

Objectivethe outcome Scopewhat it covers Requirements"shall" statements Evidencewhat proves it Metricswhat to report Basis forconclusionswhy, in public
Requirements use "shall". Guidance uses "should". The basis for conclusions explains the Board's reasoning.
Document types

Three kinds of authoritative text

Standards

Requirements an organization meets to claim conformance. Issued only after full due process and a vote of at least five of seven Board members.

Implementation Guides

Worked examples, templates and sector notes that show how to meet a standard. Non-mandatory; shorter comment period.

Interpretations

Answers to specific questions raised by members about how a requirement applies. Published with the question that prompted them.