Frequently Asked Questions
Common questions about IAIMS, our mission, and how to get involved.
What is IAIMS?
IAIMS (The Institute for AI Measurement Science) is a nonprofit organization focused on translating AI governance requirements into machine-readable evidence artifacts. We provide open, non-prescriptive infrastructure that organizations can use to demonstrate compliance with emerging regulations and voluntary frameworks.
Why does measurement matter for AI governance?
Today, most AI governance requirements are written in legal or policy prose — ambiguous, hard to operationalize, and inconsistently interpreted. IAIMS bridges this gap by creating structured mappings from regulatory text to technical evidence, helping builders prove what they've done while giving policymakers clearer visibility into what 'compliance' means in practice.
What are you currently working on?
We are currently focused on initial evidence schemas and system documentation profiles, continued framework-to-measurement mappings (EU AI Act, NIST AI RMF), and planned pilot testing with academic and practitioner partners.
Is IAIMS a standards body or a certifier?
No. We do not create new regulatory requirements or certify that organizations meet them. Our role is translation and tooling — helping interpret existing frameworks and provide open-source infrastructure for producing audit-ready evidence.
How is IAIMS funded?
IAIMS is a nonprofit operating through a combination of foundation grants, philanthropic contributions, and in-kind support. We do not charge for core artifacts or tooling, and we are committed to keeping all outputs open-source and publicly accessible.
Who governs IAIMS?
We are building a balanced board representing academia, civil society, industry, and former regulators. Our governance principles and conflict-of-interest policies are being finalized. Transparency and multi-stakeholder input are core operating values.
How can I get involved?
We welcome collaborators, especially those with expertise in regulatory interpretation, measurement science, AI/ML engineering, or open-source infrastructure. Reach out via our contact form or subscribe to our newsletter to stay informed as we open contributor pathways.
Where can I follow your progress?
We are building in public. View our to see current milestones, and we will publish updates via our newsletter, blog, and GitHub repository as artifacts become available.
How do you measure success?
We track progress through artifact adoption (are organizations using our schemas and tooling?), feedback from pilot partners and auditors, and execution against our public roadmap milestones. Success means practitioners can produce audit-ready evidence faster, and regulators have clearer visibility into what compliance actually looks like.
What is IAIMS liable for?
IAIMS provides measurement evidence—not legal guarantees or warranties about any specific AI system. Organizations remain responsible for their own compliance and deployment decisions based on our findings.
What will IAIMS NOT publish?
We do not publish proprietary training data, model architectures, or details that could be misused to game or circumvent AI safety measures. Our goal is accountability, not providing a roadmap for bad actors.
Can IAIMS measurements be wrong?
Like any measurement system, IAIMS outputs carry uncertainty. We document confidence intervals, known limitations, and version our methods so users understand what the evidence does—and doesn't—support.
Does IAIMS ingest my model data?
No. IAIMS provides measurement logic that runs locally within your infrastructure — we provide the 'tape measure,' you keep the data. The only output is the Evidence Manifest, which contains metadata, not raw model weights or personally identifiable information.
What questions does IAIMS deliberately not try to answer?
We do not determine whether an AI system is 'safe,' whether its risks are acceptable, or whether deployment should proceed. Those are judgment calls that belong to organizations, regulators, and affected communities — not to measurement infrastructure. Our role is to provide the evidence, not the verdict.
Still have questions?
Reach out to us directly and we'll get back to you within 5 business days.