Solutions · Governing AI

ISO/IEC 42001The checklist is the easy part. AIQURIS does the hard part.

The standard requires an impact assessment and a risk assessment for each AI system, and lifecycle controls that treat the risks found. It does not say how. AIQURIS does exactly this, deployment by deployment: the risks that matter, the Annex A controls that treat them, every exclusion justified, and the evidence your auditor will check.

ISO/IEC 42001 · AI system
HR screening AI assistantCandidate shortlisting · in certification scope
IMPACTMajorJob applicants affected
RISKS ASSESSED234 high
ANNEX A SAYSFOR THIS SYSTEM
A.6Verify and validate the AI system
Selection rates compared across age, gender and nationality on every release, against an agreed threshold
A.7Quality of data for AI systems
Training CVs checked to represent the applicant pool, by role and country
A.8Inform interested parties
Applicants told before screening that AI ranks them, with a route to human review
Every control is linked to the risk it treats and the evidence it requires.
The challenge

The standard says what to do. Not how.

01

The checklist is the easy part

Policy, roles, accountability, a template Statement of Applicability. Consultants, templates and platforms all cover this. It gets you through the document review, not through the audit of your AI.

02

The assessments need real depth

Each AI system needs an impact and a risk assessment. Done generically, they say nothing about the system. Done by experts, they take months per system.

03

The controls are where most fall short

Annex A says to have lifecycle controls in place. What that means for a candidate screening AI assistant, compared with a coding AI assistant, nobody tells you. The controls also cut across data, model, people and processes, and must keep up when the system changes. So organisations implement everything, or too little.

The audit stands or falls on the assessments and controls. That is where AIQURIS works.

On the road to certification

Where AIQURIS fits on the road to certification.

1

Scope and leadership

Define the scope, AI policy, roles and accountability.

You, or your adviser
2

Assess each AI system

Impact and risk assessment per deployment, at the depth its risk calls for. Annex A controls derived, every exclusion justified.

AIQURIS
3

Specify and implement controls

Your teams put the controls in place. AIQURIS specifies each one and tracks its status and evidence.

AIQURIS specifies, you implement
4

Internal audit and management review

Your management system checks itself, as the standard requires.

You
5

Certification audit

Stage 1 and stage 2 audits by an accredited certification body.

Certification body

Others give you the framework. AIQURIS fills it with your AI.

Traced, not templated

From a risk to the evidence your auditor checks.

"Have appropriate lifecycle controls in place." This is what that means for one AI system.

RISKA vendor update changes how the HR screening AI assistant ranks candidates, and nobody notices
REQUIREMENT · SOURCEManage the AI system life cycle · ISO/IEC 42001, Annex A.6
QUESTIONHow are model changes reviewed and approved before use?
EVIDENCEChange log, with re-validation results before each release
OWNERThe HR system owner approves every model change before use.

One of many chains a single assessment produces.

Why AIQURIS

Built by people who helped develop the standard.

We were involved in developing ISO/IEC 42001, and we built AIQURIS for the part the standard leaves open: assessing each AI system and specifying the controls it needs.

AIQURIS started within the global AI quality practice of TÜV SÜD, one of the world's largest testing, inspection and certification companies. Seven years of methodology development and decades of certification and safety-critical expertise stand behind every assessment. We know what an auditor looks for.

Today AIQURIS is independent and not a certification body, so we can prepare you without a conflict of interest. Certification bodies are not allowed to advise you; we are.

Meet the team and their standards work →
What you get

The assessments and controls your management system stands on.

  • ✓An impact assessment and a risk assessment for each AI system in scope
  • ✓The controls that treat each risk, with the Annex A controls that apply and every exclusion justified, in your Statement of Applicability
  • ✓Requirements, controls and evidence per deployment, linked to their source and ready for the audit
  • ✓Every AI deployment in scope, in one AI register
  • ✓Kept current as deployments change

Leadership, internal audit and management review stay with you or your adviser.

Start free

Your first impact screening (IMPACT+) is free.

Describe one AI deployment. See who could be affected, how badly, and how deep the assessment needs to go next. No sales call needed.

  • ✓Impact tier for your deployment
  • ✓Affected stakeholders and potential harms
  • ✓Where to go next, and how deep
  • ✓The deployment registered in your AI register
Start your free screening
Get started

Scope your certification on the AI you actually run.

Bring your intended certification scope and the AI systems in it. In one call we scope the deployment assessments, the Annex A decisions and the evidence your audit will need.

Privacy Overview

We use cookies to operate our website, ensure its proper functioning, improve performance, analyse traffic, and support our marketing activities. Some cookies are strictly necessary and cannot be disabled. Others can be enabled or disabled below according to your preferences. For full details, please see our Privacy Policy.