Frontier Risk·May 30, 2026·4 min read
Deployment is where risk lives.
Most of the AI risk that matters in practice doesn't come from model capability — it comes from how organizations decide to use that capability.
Read article →Independent research and advisory on AI safety
Rigorous, independent thinking on how AI is deployed in the world.
USIAIS publishes research, advises organizations, and certifies AI systems against published safety standards.
What we do
Three programs, one institutional voice.
Program 01
Research
Opinion pieces, essays, and research notes on AI safety, governance, and policy. Published openly under our editorial independence policy.
Read research →Program 02
Advisory
Monthly and weekly advisory engagements for organizations deploying AI systems. Independent perspective; consistent counsel.
See advisory tiers →Program 03
Certification
Independent assessment of AI products and organizations against published safety frameworks. Numerical scores, letter grades, twelve-month validity.
View frameworks →Latest research
What we're writing about.
Director·May 30, 2026·3 min read
From the Director: Why USIAIS exists, and what it will and won't do.
An inaugural statement on the mission, the discipline of editorial independence, and the work of bringing safety practice to the deployment of AI.
Read article →Methodology·May 30, 2026·4 min read
What an independent assessment actually looks like.
Notes on the methodology behind USIAIS Product and Company assessments, the questions they answer, and the ones they deliberately do not.
Read article →Corporate Membership
Support independent research on AI safety.
USIAIS membership is for organizations that want to contribute to independent research on AI safety and governance. Membership is not a certification, an endorsement, or an assessment. Members support our work; we maintain complete editorial independence.
Founding members
Founding Member
USIAIS-M-0001
Founding Member
USIAIS-M-0002
Founding Member
USIAIS-M-0003
Founding Member
USIAIS-M-0004
Founding Member
USIAIS-M-0005
Founding Member
USIAIS-M-0006
Founding members will be confirmed and announced prior to public launch.
Certification
An assessment is a published methodology, scored, signed, and dated.
Each USIAIS certification is anchored to a published framework, scored on a hundred-point scale, and issued with a verification number that can be checked against our public record.
Payment for an assessment does not entitle a client to a passing grade. Companies and products may, and do, receive failing grades.
View certification frameworks →U · S · I · A · I · S
Certified Organization
B
84 / 100
Strong
Acme AI, Inc.
Issued May 2026 · Valid 12 mo.
USIAIS-C-0001
From the Director
Artificial intelligence is being deployed across every industry faster than most organizations are equipped to deploy it safely. USIAIS exists to close that gap — through research, advisory work, and independent assessment. We do not believe AI is the enemy. We believe that responsible deployment is hard, that the practical work matters more than the rhetoric, and that the institutions doing this work need to be both intellectually honest and operationally useful.
— Charlie MaruccoDirector
Engage