Bias audit and notice requirements
Some jurisdictions now require an independent audit of automated employment tools, public posting of results and advance notice to candidates that a tool is being used.
Risk topic
The Software Screened Them Out. You Rejected Them.
A person gets screened out. They can challenge what happened and you may have to explain what the software did. That's why this area is moving fast.
What is already running in most hiring processes
Very few small businesses think they use AI in hiring. Almost all of them do. It came bundled with the job board, the applicant tracking system or the recruiter you hired.
The candidate never sees it. They apply, they hear nothing and a system somewhere decided they were a poor match before a person ever read a word.
- Resume parsing and keyword ranking inside applicant tracking systems
- Match scores and knockout questions on job boards
- Video interview scoring and analysis
- Personality, aptitude, and game based assessments
- Automated scheduling and chat screeners
- Background and social screening tools that summarize a candidate
Nobody asks if the tool is smart. They ask for the numbers.
Nobody has to prove you intended to discriminate. In employment, a neutral process that produces uneven outcomes is enough to start the fight and the burden then lands on you to justify the tool.
That justification has to be about the job. A vendor's validation claim won't carry it. You have to show the thing you screened on actually relates to doing the work.
A second problem sits underneath the first. Screening out someone who needed an accommodation is a separate violation with its own exposure. A timed assessment, a video interview, or a chat screener that assumes a certain way of responding can shut out a qualified person before anyone even knew they existed.
The Signal
Employment is where AI oversight has the most concrete requirements today. Much of it is local and much of it applies based on where the candidate sits regardless of where your business is located.
Automated decision rules in state privacy laws
Several state privacy laws give people rights around significant automated decisions, including the right to be told and in some cases to ask for human review.
Enforcement under existing discrimination law
Agencies have made clear that longstanding employment discrimination law applies to algorithmic screening, with no new statute required.
Accommodation and accessibility duties
Assessments and automated interviews must leave room for candidates who need an alternative and that obligation sits with the employer.
Recording and consent for video interviews
Video and voice analysis runs into recording consent and biometric rules that differ sharply between states.
Record retention
Application records, screening criteria, and the reasons for rejection carry retention obligations that most businesses satisfy by accident, if at all.
Why It Matters for your business
Hiring claims are cheap to bring and expensive to defend. The applicant already has the rejection email, the timestamp, and often the job posting that describes the tool.
Your exposure follows the candidate. A remote posting can pull applicants from states with requirements you have never read and the rule that applies is usually the one where the applicant is sitting.
Recruiters and staffing partners do not absorb this for you. In most arrangements the employer remains responsible for the decision no matter who ran the screen.
Prepare Now
- Ask your applicant tracking system and job board vendors, in writing, exactly what scoring or ranking they apply to candidates.
- Write down the job related reason for every screening criterion you use, in one sentence each.
- Make sure a person reviews rejections in a meaningful way and that the review leaves a record.
- Offer a clear alternative path for candidates who need an accommodation and put it in the posting.
- Check the notice and audit rules for every state you post into, including states outside your home state.
- Keep applications, criteria, and rejection reasons for as long as the applicable rules require.
The Evidence Test
A rejected candidate's attorney sends a letter. The question is never whether you meant well.
- What tool scored this candidate and what did it score them on?
- What is the job related justification for each criterion and who approved it?
- Has the tool been tested for uneven outcomes and can you produce the results?
- Did a human review this rejection and what shows that the review was real?
- Was the candidate told an automated tool was used and was an alternative offered?
How AI Governance Rule Watch™ keeps that record for you
When a client, insurer, or auditor asks how your business handles AI, you don't have to dig through email. You open your account and show them.
Governance Score
A quick read on where your business stands: what's missing, what's overdue, and what needs your attention next.
Living AI Tool Roster
A list of every AI tool your team uses and the rules for each one. If you stop using a tool, you keep the date, the reason, and who signed off.
Document Studio
Ready-to-adopt policies and notices, including a staff policy, client notice, vendor rider, and incident protocol. You start from a finished draft.
Check-ins and reviews
Weekly and monthly check-ins plus 3-, 6-, and 12-month Governance Reviews of your whole AI program, with reminders. When your insurer asks when you last reviewed how your business uses AI, you have the date.
Evidence Vault
A list of the records you keep, when each was last reviewed and what's due next. You never upload client files.
Diligence Ledger
A dated history of every review, decision, and policy you adopt. You can't go back and rewrite or delete completed entries and corrections keep the original history, which is why an auditor can rely on it.
One-Click AI Governance Status Report
One click produces an Auditor report and a Stakeholder report you can hand to a regulator, insurer, client, or partner the same day.
Rules watched for you
1,200+ official sources, including agencies, all 50 states, courts, and licensing boards, turned into plain-English action items.
Could you show your work tomorrow?
Hiring rules are set where your candidates live, they change every legislative session and nobody sends an employer a notice when a new one takes effect. AI Governance Rule Watch™ tracks those changes so you find out before a candidate's attorney does.
Enterprise AI governance platforms cost tens of thousands a year and are built for companies with compliance departments. A one-time memo from counsel goes stale as soon as the next rule lands. AI Governance Rule Watch™ is built for businesses like yours. We watch the official sources, tell you what to do, give you the policies, and keep a dated record of what you did. See how we compare.
Get your AI governance in order.
When an AI tool causes a problem, your client blames you. AI Governance Rule Watch™ watches the rules that apply to your business, tells you what to do about them, gives you the policies and keeps a dated record of the work. When a regulator, client, insurer, or licensing board asks, you can answer the same day.
Founding member rate. Cancel anytime. Plus applicable taxes. Renews until canceled. Not legal advice and not a compliance certification.
This guide is not legal advice. See our legal disclaimer.