3 standards bodies wrote this down separately. None of them consulted the others. That is not a preference you can argue with, it is what happens.From 3 controls across our corpus
Canada AIDA Artificial Intelligence and Data Act Evidence & Implementation Kit
This is the complete documentation set: an adopt-ready artifact for every control in policy and procedure text you edit rather than draft, and the evidence checklist an auditor asks for against each.
See what is in it, $249
The same set every buyer of this kit receives. Nothing here is produced on request.
Who warns about it
Every framework below independently names this failure. They were written by different
bodies, in different jurisdictions, for different industries, and they agree.
What an auditor asks for
The artefacts named on the controls that warn about this failure. This is what closes
it, and what you will be asked to produce when somebody checks.
- Review of models and targeting for protected characteristics and proxies
- Fairness testing results for eligibility decisions
- Legal review records for ad targeting in housing, credit and employment
- Disparate impact testing of models and targeting used for housing, employment, credit or education
- Controls on use of protected characteristics and proxies
- Review records for ad targeting in regulated sectors
What closing it also buys you
The 6 controls that warn about this failure map onto controls in other
frameworks. Close them here and this much of each of those is closed too. It is the same work
counted once, which is usually the difference between a programme that finishes and one that
does not.
ISO/IEC 29134:20233
SQF Code Edition 91
ISO/IEC 27557:20222
OECD Recommendation on Artificial Intelligence (2024 Update)3
OECD AI Principles12
ASIS SPC.1-20091
ISO/IEC 27031:20111
Scaled against the framework this reaches furthest into, not against a hundred
percent. Exact figures below.
| Also progresses | Covered |
Controls reached |
|---|
| ISO/IEC 29134:2023 International (ISO/IEC JTC 1/SC 27); adopted as CSA ISO/IEC 29134:24 and nationally | 3.5% | 3 of 85 |
| SQF Code Edition 9 International (GFSI-benchmarked; SQFI, a division of FMI, Arlington, VA) | 1.8% | 2 of 113 |
| ISO/IEC 27557:2022 International (ISO/IEC JTC 1/SC 27) | 2.6% | 2 of 77 |
| OECD Recommendation on Artificial Intelligence (2024 Update) International (OECD; 51 adherents per the Compendium record of 2026-09-25) | 3.0% | 1 of 33 |
| OECD AI Principles International | 12.5% | 1 of 8 |
| ASIS SPC.1-2009 United States (ANSI-accredited standard, used internationally; adopted under DHS PS-Prep in 2010) | 1.3% | 1 of 77 |
| ISO/IEC 29147:2018 International (ISO/IEC JTC 1/SC 27); adopted as EN ISO/IEC 29147:2020 and nationally | 0.9% | 1 of 113 |
| ISO/IEC 27031:2011 International | 1.1% | 1 of 90 |
Read as: closing this failure reaches that share of the named
framework's control library through cross-framework mappings held in our corpus. It is not a
claim of compliance with that framework, it is a measure of how much of it you have already
touched.
Where this comes from
Harvested from the control library itself. Every control in our corpus carries the evidence an
auditor expects and the ways implementations commonly fail, recorded when that control was
verified against its source document. This page is those two fields, for one failure, across
every framework that names it.
The overlap is computed by traversing
332,959 cross-framework control mappings out from the specific controls that
warn about this failure, not from the frameworks they sit in. Those mappings were built control
by control against source documents.
Nothing here is inferred, predicted or scored. The number at the top
is a count of frameworks.
All failure modes ยท
The index