Not legal advice. A plain-English map of the laws that apply when a trade business uses AI or automated software to hire, dispatch, score, or discipline technicians. Current as of November 2026. Several of these laws are new, have pending rules, or are in litigation, so confirm with an employment attorney before you rely on any of it.

Quick answer

There’s no federal AI hiring law, but federal discrimination law still applies to AI decisions, and a growing list of states and cities now regulate automated hiring and management tools directly. For a trade shop, the risk isn’t a robot hiring your techs. It’s a screener, scorecard, or auto-dispatch feature quietly making decisions nobody reviews.

  • Hiring: NYC requires annual bias audits and 10 business days’ notice. Illinois requires notice and bans ZIP-code proxies. New Jersey codified disparate-impact liability for automated tools.
  • Coming January 1, 2027: Colorado’s rewritten AI law and California’s CCPA rules on automated decision-making, which explicitly cover “allocation or assignment of work.”
  • July 1, 2027: California’s No Robo Bosses Act bars firing or disciplining based solely on software.
  • October 1, 2027: Connecticut’s pre-decision notice for automated employment tools, though scheduling and productivity monitoring are excluded.
  • Safe default everywhere: tell people when software is involved, keep a human making the final call, and make your vendor explain what the tool does.

Most trade owners don’t think they use AI in hiring. Then you look at the stack: a job board that ranks applicants by “match score,” a phone agent that screens candidates after hours, a dispatch board that auto-assigns calls, a scorecard that flags techs with high callback rates, and a dashcam that grades driving. Each of those makes or shapes a decision about a person’s job. That’s exactly what the new laws are written to reach.

This guide sorts the laws by what they actually touch in a service business: who you hire, and how you manage the techs you have.

Where AI hides in a trade shop

ToolWhat it doesLaws that may care
Job board ranking and “match scores”Sorts or hides applicants based on resume analysisNYC LL 144, Illinois HB 3773, New Jersey rules, California FEHA regs, Colorado, Title VII and ADEA
Automated knockout questionsRejects applicants who answer “no” to a screenerFederal disparate impact; Illinois if criteria act as proxies
AI phone or chat screenersInterviews and scores candidates, books interviewsIllinois AI Video Interview Act (video), Maryland (facial recognition), NYC, Colorado, Connecticut
Auto-dispatch and job assignmentAssigns calls by location, skill, or performanceCalifornia CCPA ADMT rules (“allocation or assignment of work”), possibly Colorado and Illinois
Tech scorecards and rankingsRanks techs on callbacks, close rate, reviews, drive timeCalifornia SB 947, Illinois, Colorado, Connecticut (if used for discipline or pay)
AI dashcams and driver scoringGrades driving, sometimes analyzes facesCalifornia AB 1883 and SB 947; state notice laws
Commission and pay automationCalculates pay from job dataWage law as usual; Colorado if the tool decides compensation

A simple calculation, like a commission split your rules define, is different from a tool that predicts, scores, or ranks people. Most of these laws target the second kind. Definitions vary by law.

The laws at a glance

JurisdictionLawEffectiveWhat it requiresEnforcement
New York CityLocal Law 144 (AEDT)Enforced since July 2023Independent bias audit within a year before use, public audit summary, notice to candidates and employees 10 business days before useCivil penalties, $500 first violation, up to $1,500 after
IllinoisHB 3773, amending the Human Rights ActJan. 1, 2026No AI use that has a discriminatory effect; no ZIP codes as a proxy; notice when AI is used in recruiting, hiring, discipline, discharge, and other terms of employmentIllinois Human Rights Act process; notice rules proposed and withdrawn in 2026
IllinoisAI Video Interview Act (820 ILCS 42)2020Notice, explanation, and consent before AI analyzes video interviews; limits on sharing; deletion on requestStatutory
MarylandHB 12022020No facial recognition templates during interviews without a signed consent waiverStatutory
New JerseyN.J.A.C. 13:16 (Division on Civil Rights)Dec. 15, 2025Codifies disparate-impact liability under the Law Against Discrimination, naming automated decision toolsCivil rights enforcement; employer liable for vendor tools
CaliforniaFEHA automated-decision system regulationsOct. 1, 2025Discrimination law applies to automated decisions; anti-bias testing (or the lack of it) is relevant evidenceCivil Rights Department and lawsuits
CaliforniaCCPA automated decision-making technology (ADMT) regulationsJan. 1, 2027 for existing usesPre-use notice, opt-out (with exceptions), access rights, and risk assessments for “significant decisions,” including hiring, pay, and allocation of workCPPA and Attorney General, $2,500 per violation, $7,500 intentional; only businesses covered by the CCPA
CaliforniaSB 947, No Robo Bosses ActJuly 1, 2027No sole reliance on automated systems for discipline or termination; human corroboration; post-use noticeLabor Commissioner, $500 per violation
ColoradoSB 26-189 (replaced SB 24-205)Jan. 1, 2027; enforcement tied to AG rulemakingNotice before automated tech materially influences a consequential decision; plain-language explanation after an adverse outcome; data correction; human review where commercially reasonable; 3-year recordsAttorney General only, up to $20,000 per violation
ConnecticutSB 5, Public Act 26-15Oct. 1, 2026 (WARN, discrimination defense); Oct. 1, 2027 (notice)Pre-decision written notice with the tool’s trade name, purpose, and data used; AI is not a defense to discriminationAttorney General, 60-day cure through 2027
TexasTRAIGA, HB 149Jan. 1, 2026No AI deployed with intent to discriminate; disparate impact alone isn’t enough. Most duties fall on governmentAttorney General

Sources: Epstein Becker Green; Cooley; Akin Gump; Rocky Mountain Employer blog; AI Compliance Atlas; SiftFirst tracker; Proskauer; Ropes & Gray. Checked November 2026. NYC penalty amounts are from the city’s rules as commonly reported; verify with NYC DCWP.

The federal picture

No federal statute regulates AI hiring tools specifically. But the old laws still apply to new tools:

Hiring laws in detail

New York City

If you hire for jobs in NYC and use an automated employment decision tool to “substantially assist” hiring or promotion, you need an independent bias audit within the year before use, a public summary of the results, and notice to candidates at least 10 business days before the tool is used. There’s no small-business exemption. Most job boards’ built-in ranking is the gray area; ask the vendor whether they’ve had an audit done and whether it covers how you use the tool.

Illinois

Illinois has two laws. HB 3773, effective January 1, 2026, amends the Illinois Human Rights Act. It applies to employers with one or more employees and covers AI used in recruiting, hiring, promotion, training selection, discipline, discharge, tenure, and the terms of employment. You can’t use AI in a way that has a discriminatory effect, you can’t use ZIP codes as a proxy for protected classes, and you have to give notice. The Department of Human Rights proposed notice rules in May 2026 and withdrew them in June, so there’s no official template yet. Give notice anyway.

The ZIP code ban matters to trade shops. “Only show me applicants within 15 miles” feels like a dispatch decision. Depending on how it’s implemented and the area, it can screen out neighborhoods along racial lines. Base distance rules on actual job requirements, like on-call response time, and document why.

The AI Video Interview Act applies if AI analyzes recorded video interviews: you have to notify the applicant, explain how the AI works and what it evaluates, get consent, limit sharing, and delete the video on request.

New Jersey, California, Maryland, and Texas

Dispatch, scoring, and discipline

This is where trade shops are more exposed than office employers, because dispatch software touches pay and performance every day.

California: two separate laws

The CCPA’s ADMT regulations cover technology that processes personal information and uses computation to replace human decision-making in “significant decisions.” For employment, that list includes hiring, compensation, promotion, demotion, suspension, termination, and, notably, “allocation or assignment of work.” Businesses already using ADMT have until January 1, 2027 to provide a pre-use notice, an opt-out or a qualifying exception, access rights, and a risk assessment. This only applies to businesses covered by the CCPA, which generally means about $26.6 million or more in annual revenue or large-scale data handling. Most small shops aren’t covered; a big multi-location HVAC company may be.

SB 947, the No Robo Bosses Act, applies to all California employers starting July 1, 2027. You can’t rely solely on an automated decision system to discipline or fire someone. If you primarily rely on one, a human must independently corroborate the decision using the same data, and you have to give the worker a written notice when you tell them the decision. See California employment laws for trade employers.

Colorado

SB 26-189, signed May 14, 2026, repealed and replaced Colorado’s original AI Act. It drops the original law’s risk management program and annual impact assessments. Starting January 1, 2027, with enforcement tied to the Attorney General finishing rules, a business that uses automated decision-making technology to materially influence a “consequential decision” must:

A consequential decision includes one related to access to, eligibility for, or compensation related to employment. For a commission tech, a tool that decides who gets the high-ticket install calls is arguably a compensation decision. That question hasn’t been answered. Enforcement of the original law is also tied up in a federal lawsuit filed by xAI in April 2026, with the parties agreeing to a temporary stay.

Connecticut

Connecticut’s SB 5 is narrower for dispatch. Its pre-decision notice requirement, starting October 1, 2027, covers hiring, promotion, discipline, discharge, training selection, and other material terms of employment. It expressly excludes nonmaterial changes to tasks, hours, or assignments, and decisions about health and safety, scheduling and planning, and productivity monitoring. Routine dispatch is likely out; using a scorecard to discipline or fire is in. Starting October 1, 2026, using AI is also no defense to a discrimination claim.

Illinois

HB 3773’s coverage of “terms, privileges, or conditions of employment” is broad enough that a tool that affects techs’ work and pay could fall under it. Give notice.

Five real trade-shop scenarios

2

An AI phone agent screens applicants after hours and scores them

This is an automated employment decision tool in plain terms. In NYC, it likely needs a bias audit and notice. In Illinois, notice, and if it analyzes video, consent. In Colorado starting 2027 and Connecticut starting October 2027, notice. Fix: have the agent collect information and book interviews, and have a person decide who moves forward. Tell applicants it’s an automated system.

3

Auto-dispatch sends the big install jobs to the top-ranked tech

If techs are on commission, the assignment rule moves money. California’s CCPA rules name “allocation or assignment of work” as a significant decision for covered businesses. Colorado covers decisions about compensation related to employment. Connecticut excludes scheduling and nonmaterial assignments. Fix: keep a dispatcher able to override, write down the assignment rules, and make sure the ranking isn’t built on anything that tracks age, disability, or other protected traits.

4

The scorecard flags a tech, and HR writes them up automatically

This is what California’s SB 947 targets starting July 1, 2027, and it’s covered by Connecticut’s notice rules from October 2027 and Illinois today. Callback rates can be skewed by things the tech doesn’t control, like being sent to older equipment. A tech on intermittent medical leave may look “low production.” Fix: scores start a conversation; a manager reviews the jobs behind them and documents the decision.

5

An AI dashcam scores drivers and triggers discipline

In California, AB 1883 bars AI surveillance that infers emotional state, starting January 1, 2027, and SB 947 bars sole reliance on the score for discipline. Other states have notice rules for monitoring. Fix: ask the vendor exactly what the camera infers, turn off emotion or attention inference where you can, and have a manager review the footage before acting. See GPS tracking laws for technicians.

Eight questions for your software vendor

Most of these laws put the obligation on you, the employer, not the vendor. Send these in writing and keep the answers:

1Does any feature we use rank, score, predict, or filter people, or does it just follow rules we set?
2What data does it use, and does it use anything that could stand in for age, race, disability, or ZIP code?
3Has it had an independent bias audit? Can we see the summary, and does it cover how we use the tool?
4Can a human override every automated decision, and is the override logged?
5Can we turn off automated rejection, ranking, or assignment and keep the rest?
6Can you give us a plain-language description of how it works for our notices?
7Does any camera or monitoring feature infer emotion, fatigue, or attention?
8How long do you keep the data, and can we export it for three-year recordkeeping?

A plain-language notice you can adapt

There’s no single notice that satisfies every jurisdiction, and Colorado, Connecticut, and California each spell out required content. But a clear notice is the common thread. Have your attorney adapt this starting point:

“[Company] uses software to help with [screening job applications / assigning service calls / reviewing performance data]. The software is [tool name] from [vendor]. It looks at [the information you provide in your application / job history, callbacks, and customer reviews] to [rank applicants / suggest assignments / flag performance trends]. A [hiring manager / dispatcher / service manager] reviews the results and makes the final decision. You can ask what information was used, correct anything that’s wrong, and ask a person to review a decision by contacting [name, email, phone]. We won’t retaliate against you for asking.”

Checklist

☐List every tool that touches hiring, assignment, pay, performance, or discipline.
☐Mark which ones rank, score, or predict versus just follow your rules.
☐Map where your applicants and techs work: NYC, Illinois, New Jersey, California, Colorado, Connecticut.
☐Send the eight vendor questions and keep the answers.
☐Put a human decision-maker on every hire, discipline, and firing, and document what they reviewed.
☐Remove screening criteria you can’t tie to the job.
☐Post and send notices where required, and use one everywhere as a best practice.
☐Keep records for at least three years.
☐Calendar January 1, 2027 (Colorado, California CCPA), July 1, 2027 (SB 947), and October 1, 2027 (Connecticut).

The simplest compliance strategy is also the best management: software organizes the information, and a person makes the call. VortechPro keeps every job on record by tech, so the person making the call can see what’s behind the numbers. $99/mo for 5 techs, +$20 each.

Dispatch with a human in charge

VortechPro: dispatch and job records by tech. $99/mo for 5 techs, +$20 each. 30-day free trial, no contract.

FAQ

Is there a federal law on AI in hiring?
No federal statute regulates AI hiring tools specifically, but Title VII, the Age Discrimination in Employment Act, and the Americans with Disabilities Act apply to decisions made with software. In 2023, the EEOC settled with iTutorGroup for $365,000 over software that automatically rejected older applicants.
Which states regulate AI in hiring?
As of November 2026, notable laws include New York City Local Law 144, Illinois HB 3773 and the AI Video Interview Act, Maryland's facial recognition interview law, New Jersey's disparate-impact rules, California's FEHA automated-decision regulations and CCPA ADMT rules, Colorado SB 26-189, Connecticut SB 5, and Texas TRAIGA.
Does the Trump executive order preempt state AI laws?
Not on its own. Executive Order 14365 created a DOJ task force to challenge state AI laws, but an executive order lacks preemptive force by itself. Until a court invalidates a state law or Congress passes a preemption statute, employers must comply with state AI laws.
Does auto-dispatch software fall under AI employment laws?
It can. California's CCPA automated decision-making rules list allocation or assignment of work as a significant decision for covered businesses, and Colorado covers decisions about compensation related to employment. Connecticut's law excludes scheduling, planning, and nonmaterial assignment changes.
What does the Colorado AI law require from employers?
Under SB 26-189, effective January 1, 2027 with enforcement tied to Attorney General rulemaking, businesses using automated decision-making technology to materially influence consequential decisions, including employment, must give notice, explain adverse outcomes, allow data correction and human review where commercially reasonable, and keep records for three years. Penalties run up to $20,000 per violation.
What is NYC Local Law 144?
NYC's law on automated employment decision tools requires an independent bias audit within one year before use, a public summary of the audit, and notice to candidates and employees at least 10 business days before the tool is used for hiring or promotion decisions in New York City.
Can I use a dispatch scorecard to discipline technicians in California?
Starting July 1, 2027, California's No Robo Bosses Act bars relying solely on an automated decision system for discipline or termination. If you primarily rely on one, a human must independently corroborate the decision, and you must give the worker a written notice.
Are employers liable for discrimination by their AI vendor's tool?
Generally yes. Employers are responsible for employment decisions made with vendor tools, New Jersey's rules state employers can't outsource that responsibility, and in Mobley v. Workday a federal court let age discrimination claims against the vendor itself proceed as a collective action.