Scoring or ranking people
Credit, fit, fraud, severity, applicant rank, claim priority — any number that orders human outcomes.
CCPA §7220 · CO AI Act §6-1-1701
Is your AI compliant?
If your AI influences decisions about people — hiring, lending, insurance, housing — Automated Decision-Making Technology (ADMT) regulations already apply. Most companies don't know they're in scope.
Compliance firms charge $5,000+ for this. Five minutes, no cost.
THE ADMT ENFORCEMENT TIMELINE
Here's what's already enforced — and what's next.
$4M+ in CPPA settlements since January 2026. Hundreds of open investigations.
LIVECPRA ADMT risk assessmentsIf your AI scores, ranks, or filters people, you need a documented risk assessment now — or you're already exposed.
LIVEIllinois AI hiring law (HB 3773)AI in hiring, promotion, or termination without notice and bias testing is a civil rights violation. Penalties up to $100K.
JUN 2026Colorado AI ActHigh-risk AI deployers must document impact assessments and risk management programs. $20,000 per violation.
AUG 2026EU AI Act Article 12 loggingHigh-risk AI systems need automatic event logging from day one. No retroactive fix. Up to 3% of global revenue.
JAN 2027CPRA ADMT consumer rights (§7220-7222)Your customers can demand AI decision explanations, opt out of automated decisions, and request human review.
APR 2028CPPA assessment submission deadlineEvery risk assessment you should be building now gets submitted to the state for review.
$7,500Per consumer. Per violation. No cap.10,000 affected customers × one violation = $75M exposure.
Which of these apply to your AI? Find out in 5 minutes →
WHERE ADMT REGULATIONS APPLY
ADMT regulations don't target vendors. They target actions.
If your AI does any of these — regardless of which platform you built it on — you're in scope.
Filtering before a human looks
Résumé screeners, claim auto-routing, eligibility cutoffs. If the AI says “no” and a person never sees it, that’s an automated decision.
NYC Local Law 144 · EEOC 2024 guidance
Pricing, approval, or denial
Premiums, loan terms, plan eligibility, dynamic pricing on personal data — outputs that change what a person pays or receives.
CCPA ADMT · CO SB21-169
Recommendations humans rarely override
Even with a human in the loop, if the override rate is near zero the regulators treat it as automated. The “rubber-stamp” exception.
CCPA §7001(e)
HOW IT WORKS
What if your AI didn't trigger ADMT at all?
If it can't decide, it's not in scope
ADMT regulations target automated decisions. If your AI is architecturally advisory-only, the regulations don't trigger. That's the legal standard — we make it provable.
We make that true
Every AI recommendation requires documented human review before it becomes a decision. Cryptographic proof, not just policy.
The risk assessment, ready when you need it
Generated from your audit trail. Signed records, dwell-time evidence, three-prong proof. You submit it. Done.
Compliance Dashboard
Mar 21, 2026Total Decisions
1,515
+12.3%Override Rate
14.1%
+2.1%Avg Review Time
3m 07s
-8.2%Pending Reviews
7
-3Override rate spike detectedHiring decisions — 23% overrides in last 24h (baseline: 14%)
2h agoReview time below threshold3 decisions completed in under 60s — may indicate rubber-stamping
45m agoDecision Volume (7d)
Action Items
highReview 3 flagged hiring overridesAK
highInvestigate rubber-stamp alertsMR
mediumUpdate review criteria for claimsAK
lowSchedule quarterly audit reviewJL
THE DIFFERENCE
Other platforms give you a checklist. You still have to do the work.
Elsewhere
- A checklist of what you need to fix
- A 6-month project you staff yourself
- Your AI can still bypass human review
With us
- Enforcement shipped into your code
- Deployed in weeks, not quarters
- Your AI physically cannot decide without human review
COMMON QUESTIONS
Before you ask.
What is ADMT?
Automated Decision-Making Technology. It's the regulatory term for any AI that influences decisions about people — hiring, lending, insurance, housing. California, Colorado, Illinois, and the EU all have laws targeting it, with fines already being issued.
Our AI just makes recommendations. Are we in scope?
If your approval rate is 98% with 3-second review times, your AI is the de facto decision-maker. We prove meaningful human review with cryptographic evidence — or help you close the gap before it matters.
What happens after the scan?
The scan identifies which ADMT regulations apply and where you're exposed. If they do, we deploy enforcement that makes your AI architecturally incapable of deciding without human review. Scan is free. Implementation is a paid engagement.
We already use Vanta. Why do we need this?
Vanta covers SOC 2, ISO 27001, HIPAA — none address automated decision-making. ADMT requires per-decision audit trails, advisory-only enforcement, and consumer opt-out mechanisms. Different domain entirely.
How is this different from Credo AI or OneTrust?
They cover broad AI governance. We cover ADMT only — deeply. They document compliance posture. We enforce it architecturally. Deployed in weeks, not quarters.
Does this apply outside California?
Yes. Colorado's AI Act (Jun 2026) carries $20K per violation. Illinois HB 3773 covers AI in hiring. The EU AI Act (Aug 2026) penalizes up to 7% of global revenue. If you serve customers or make decisions about people in any of these jurisdictions, you're in scope.
What does the scan actually look at?
Our AI agent researches your product, identifies every automated decision that touches a person, maps each one to the regulations that apply, and scores your compliance exposure. It's the same analysis a compliance consultant would do — delivered in minutes instead of weeks.
GET STARTED
Find out if you're in scope — then close the gap
Hiring · Lending · Insurance · Healthcare · Housing · Education
ADMT Risk Scan
Our AI agent scans your product and tells you whether ADMT regulations apply — and where you're exposed. Free.
No sign-up required. Results in minutes.
or
Talk to us
Walk through your AI systems with us. We'll map your exposure and tell you exactly what you need to submit.
Book a Call →