What is AI for recruiting when you don't have a talent team?
AI for recruiting is software that sources candidates, screens resumes, schedules interviews, and scores structured interview answers — jobs a dedicated talent team would otherwise own. Small companies are catching up fast: AI adoption in HR tasks climbed to 43% in 2025, up from the 26% reported by HR professionals in 2024 (SHRM, 2025 Talent Trends). For a company with no in-house recruiter, the right AI stack does not replace judgment — it replaces the hours a founder or office manager currently loses per open role to resume triage and scheduling back-and-forth.
Recruiting is where that adoption concentrates: among organizations using AI in HR, nearly two-thirds (64%) apply it to recruiting, interviewing, and hiring. But a 20-person company using an AI screening tool carries the same Title VII exposure as a Fortune 500 talent-acquisition department, minus the in-house counsel that would normally catch a problem before it becomes a charge. One threshold matters here: Title VII applies to employers with 15 or more employees, so a five-person company sits outside it — while state law, which often starts at one employee, does not wait for you to reach fifteen.
Small companies are usually not the ones building AI hiring models — they are buying tools that were built for enterprise talent teams and applying them without a compliance layer. That is the actual risk. Not the AI. The absence of the process that was supposed to wrap around it.
Why does a company with no recruiter need AI for hiring at all?
A company without a dedicated recruiter needs AI for hiring because the alternative is a founder, office manager, or hiring manager doing sourcing and screening manually on top of their real job — and that person quits triage before the process improves. AI compresses the repeatable steps (sourcing, first-pass resume screening, interview scheduling) so a human spends time only where judgment is non-substitutable: the actual interview and the final call.
This is not a hypothetical gap. SHRM reports that 89% of HR professionals whose organization uses AI in recruiting say it saves time or increases efficiency (SHRM, 2025), and the most common uses — writing job descriptions (66%) and screening resumes (44%) — are precisely the tasks that eat a non-recruiter's calendar. For a five-person ops team hiring its first sales rep, that is the difference between filling the role in three weeks or three months.
The mistake small companies make is treating "AI for recruiting" as one product. It is four distinct jobs — sourcing, screening, scheduling, scoring — and each carries a different level of legal exposure. Bundling them into a single "hire the AI recruiter" decision is how compliance gaps get created without anyone noticing.
Where does the legal exposure actually sit in the hiring funnel?
The legal exposure sits almost entirely in screening and scoring, not sourcing or scheduling — anywhere the software narrows or ranks a candidate pool. Screening tools that filter resumes or score interviews against a model trained on prior "successful" hires are the parts regulators have written rules around by name.
| Recruiting stage | Where AI helps | Legal exposure | Required human decision point |
|---|---|---|---|
| Sourcing (finding candidates) | AI scans job boards, LinkedIn, internal databases to surface candidates matching a role | Low — NYC DCWP states expressly that using a tool to scan a resume bank, conduct outreach to potential candidates, or invite applications does not trigger Local Law 144's requirements | Recruiter/hiring manager sets and reviews search criteria to avoid proxy discrimination |
| Screening (resume/application filtering) | AI ranks or filters applicants against job criteria | High — this is the "substantially assists or replaces discretionary decision-making" trigger under NYC Local Law 144 if hiring for an NYC-based role, and a "selection procedure" under the federal Uniform Guidelines | Human reviews the filtered-out pool before rejection is finalized; annual bias audit by an independent auditor required if an AEDT is used for NYC-based roles |
| Interview scheduling | AI coordinates calendars, sends reminders, reschedules | Minimal — logistics only | None required, but candidates should be told AI is handling scheduling |
| Structured interview scoring (incl. video) | AI scores recorded or live interview responses against a rubric | Highest — a "selection procedure" under 29 CFR 1607.3, and directly implicated by Illinois' HB 3773, effective January 1, 2026 | A human makes the actual pass/fail or advance/reject call; the AI score is an input, never the decision |
| Offer / final decision | AI can draft offer letters; should not weight the go/no-go | High if AI output is treated as the decision itself | Hiring manager makes and documents the final call independent of any AI ranking |
The pattern across every rule that has actually taken effect: none of them ban AI in hiring. All of them require that a human remain the decision-maker and that the employer can explain, on request, why the tool ranked candidates the way it did.
What does NYC Local Law 144 actually require, and does it apply outside New York City?
NYC Local Law 144 applies only where an Automated Employment Decision Tool (AEDT) is used "in the city." Per NYC's Department of Consumer and Worker Protection, that means the job location is an office in NYC at least part time, or the job is fully remote but the location associated with it is an NYC office — so a small business hiring for a Raleigh or Apex role is generally outside its reach.
Where it does apply, the law requires a bias audit completed before use and within the past year, performed by an independent auditor, whose evaluation must at minimum calculate selection or scoring rates and the impact ratio across sex categories, race/ethnicity categories, and intersectional categories. Employers must publish a summary of the most recent audit and the tool's distribution date, and must provide notice to NYC-resident candidates 10 business days before using an AEDT.
Two details matter more than the rest for a small company. First, "employment decision" is defined broadly to include screening — not just the final hire. Second, and critically, DCWP states that the requirements do not apply when a tool is used to scan a resume bank, conduct outreach, or invite applications; they attach only when assessing candidates who have applied for a specific position. That is the clearest official statement anywhere that sourcing and screening are legally different activities.
If you are not hiring for NYC-based roles, Local Law 144 is not a compliance requirement — but it is the closest thing the country has to a proven bias-audit template, and adopting its structure voluntarily is cheap insurance against the federal standard below, which applies everywhere.
What federal rules apply to AI hiring tools everywhere in the US?
The operative federal rules are Title VII and the Uniform Guidelines on Employee Selection Procedures (29 CFR Part 1607), which apply to algorithmic tools because of how broadly they define what counts as a selection procedure. Under 29 CFR 1607.16, a "selection procedure" is "any measure, combination of measures, or procedure used as a basis for any employment decision" — language that comfortably covers a resume screener or an interview-scoring model.
The operative standard is unambiguous:
"The use of any selection procedure which has an adverse impact on the hiring, promotion, or other employment or membership opportunities of members of any race, sex, or ethnic group will be considered to be discriminatory and inconsistent with these guidelines, unless the procedure has been validated in accordance with these guidelines, or the provisions of section 6 below are satisfied." — 29 CFR 1607.3(A)
The same regulation answers the vendor question. A "user" is defined as "any employer, labor organization, employment agency, or licensing or certification board … which uses a selection procedure as a basis for any employment decision." The employer is the user. Buying the tool from a vendor does not move that obligation onto the vendor — which is why the practical step, before turning on any AI screening or scoring feature, is to ask the vendor in writing what adverse-impact testing they have run, and to keep the answer.
One housekeeping note worth knowing, because a lot of published advice still points at it: the EEOC's 2023 technical-assistance document on assessing adverse impact in AI selection procedures is no longer available on eeoc.gov, and the link to it from EEOC's own publications index now returns a 404, as does its companion ADA guidance. Treat any article that cites those documents as the current federal standard with caution. Part 1607, by contrast, remains in the Code of Federal Regulations, current as of August 2026.
What does Illinois' AI employment law require, and who does it cover?
Illinois House Bill 3773 amended the Illinois Human Rights Act effective January 1, 2026, and expressly prohibits employers from using AI that "has the effect of subjecting employees to discrimination on the basis of protected classes" with respect to recruitment, hiring, promotion, renewal of employment, selection for training or apprenticeship, discharge, discipline, tenure, or the terms, privileges, or conditions of employment (National Law Review / K&L Gates).
HB 3773 also requires employers to notify employees and applicants when AI is used across that same list of decisions. Practically, that means an Illinois employer has to understand well enough how its AI hiring tools work to explain their use in plain language — which is a meaningfully higher bar than clicking through a vendor's setup wizard. Counsel for K&L Gates note that although the bill does not use the word "applicants" explicitly, the IHRA extends to job applicants and the law's requirements reach "recruitment" and "hiring," so employers should comply with respect to applicants too.
Where does Colorado's AI law currently stand for employers?
Colorado's AI Act has not taken effect and its status changed repeatedly — small companies hiring in Colorado should not build compliance programs around the original statute. SB 24-205, signed in 2024, would have regulated "high-risk" AI used in consequential decisions including employment, requiring impact assessments, disclosures, and risk management programs, with fines up to $20,000 per violation. Its original February 1, 2026 effective date was pushed to June 30, 2026 by a special-session bill passed August 26, 2025 (Clark Hill).
Before that date arrived, the framework was replaced. After xAI sued and the DOJ intervened, a federal magistrate judge issued an order on April 27, 2026 halting enforcement, and on May 14, 2026 SB 26-189 was signed into law, repealing and reenacting SB 205 with a narrower, notice-based approach and pushing the effective date to January 1, 2027 (McDermott Will & Schulte). Notably, the replacement drops SB 205's affirmative algorithmic-discrimination obligations but retains consumer-facing notice, post-adverse-outcome explanation, and human review.
For a small business, the operational takeaway is not "ignore Colorado" — it is that human review and disclosure survived even the deregulatory rewrite, so building toward that bar now avoids a scramble in 2027.
How should a small company without an HR team actually deploy AI in hiring?
A small company should deploy AI for the low-exposure stages first — sourcing and scheduling — before touching screening or scoring, and should never let an AI score be the final word on a candidate. That single rule satisfies most of what NYC, Illinois, and the federal Uniform Guidelines ask for simultaneously: a documented human decision point at the moment the AI's output could change someone's employment outcome.
In practice that means: use AI to find and long-list candidates freely — the stage NYC regulators have expressly carved out. Use AI to draft job posts and coordinate interview logistics freely. When AI screens resumes or scores structured interview answers, route every rejection through a human who reviews the borderline cases before the AI's filter becomes final — and keep a record of that review. That record is the artifact that turns "we used a vendor's AI tool" into a defensible process if a candidate ever challenges the outcome.
This is the same infrastructure logic isonew applies to every GTM system: the AI does the repeatable work, a human owns the decision, and the company — not a vendor's black box — holds the record of why a candidate advanced or didn't. See isonew's broader AI workflow automation guide for small business for how this same human-checkpoint pattern applies outside of hiring, and the small business AI toolkit for tool selection criteria that apply to recruiting software too.
For the full picture of how small businesses are deploying AI across every function — not just recruiting — see isonew's AI for small business hub.