AI for staffing agencies: where it helps recruiters, and the hiring rules that apply

How do staffing agencies use AI today?

You may already use AI without calling it that: the resume search in your applicant tracking system, the suggested replies in your inbox, the job board that sorts who applied. What has changed is how much of the writing and sorting software can now take on.

Recent survey numbers cover HR teams across employers, with no separate figure for staffing agencies. SHRM research reported in April 2026, drawn from 1,908 HR professionals, found:

• 39% of organizations use AI in their HR functions, and another 23% use it elsewhere in the business.
• 27% of organizations apply AI in talent acquisition, which SHRM calls the clear leader among HR functions.
• The common recruiting uses are writing job postings, screening and matching resumes, and scheduling interviews.

AI in staffing covers the same uses. An agency recruits for many clients at once, so posting, searching and scheduling come round every day.

How do I use AI for recruiting?

Start with the work that is writing or admin, and keep the judgment with your recruiters. AI for recruiting is useful for:

• Job posts and outreach: a first draft of the ad from the client’s job order, and outreach messages a recruiter edits before sending.
• Searching your own database: past candidates who match a new job order by skills, location and availability, before you pay to post.
• Summaries: a resume summarized against the job’s requirements, or a phone screen summarized from the recruiter’s notes.
• Scheduling: interview times offered from recruiters’ and hiring managers’ calendars, with confirmations sent.
• Clean records: duplicate candidate records merged, missing fields filled and stale profiles flagged.
• Paperwork: details pulled from job orders, timesheets and onboarding forms into your systems.

ChatGPT for recruiting fits the drafting and summary work well, on a business account whose data terms you have checked. Pasting a candidate’s personal details into someone’s personal account puts them outside your agency’s control.

Which recruiting tasks can AI take on, and which stay with a recruiter?

A useful line runs between tasks that prepare a decision and tasks that make one:

Only one row there makes a decision about a person, and it is the row most of the rules below are written for. Discrimination law still applies to every row.

What hiring laws apply to AI in recruitment?

None of the rules below bans AI from hiring. They say what you must check, publish, tell candidates and keep when software helps decide who moves forward. Four are worth knowing:

• New York City, Local Law 144: an employer or employment agency can’t use an automated employment decision tool unless the tool has had a bias audit within the past year, a summary of the results is public, and candidates are given notice. The city defines such a tool as one that uses machine learning, statistical modeling, data analytics or AI and substantially assists or replaces discretionary decisions about hiring or promotion. Notice goes out 10 business days before use, and enforcement began on July 5, 2023.
• Illinois, HB 3773: from January 1, 2026, the Illinois Human Rights Act bars using AI in a way that results in discrimination at work, with AI defined broadly enough to include generative AI. The law also calls for notice when AI is used, and in April 2026 the details of that notice were still to come. Check where the notice rules stand before you rely on them.
• California, Civil Rights Council: regulations in effect from October 1, 2025 say that using an automated-decision system may violate California law when it harms applicants or employees on a protected basis, and they require automated-decision data to be kept for at least four years.
• California, Privacy Protection Agency: from January 1, 2027, businesses subject to the California Consumer Privacy Act must follow the agency’s rules when they use automated decisionmaking technology for significant decisions. In work, those decisions include hiring, assigning work and setting pay, promotion, and demotion, suspension or termination. A tool falls in scope when its output is used to make the decision without human involvement, which the rules define as a reviewer who knows how to read the output, weighs it with the other relevant information and has the authority to make or change the decision.

New York’s audit compares a tool’s selection or scoring rates across sex and race or ethnicity categories. The city’s FAQ says the law doesn’t require any specific action based on the results, though federal, state and city discrimination laws still apply to what you do with them.

Enforcement was light in its first two years. In December 2025, New York State’s Comptroller reported that the city’s Department of Consumer and Worker Protection received two complaints about these tools from July 2023 to June 2025. The department had checked the websites and bias audits of 32 companies and found one issue of non-compliance. The Comptroller’s auditors reviewed the same companies, found at least 17 instances of potential non-compliance and called the complaint process ineffective. Complaints can still be filed: the city takes reports through 311 and its website, and it refers discrimination claims to its Commission on Human Rights. This section is a summary, so talk to your employment lawyer before a screening tool goes in front of candidates.

Can ChatGPT screen resumes?

It can read a resume and say how it compares with a job’s requirements. Whether it should be the thing that decides is a separate question, for three reasons:

• It can be wrong with confidence: a language model can misread dates, miss a qualification or describe experience the candidate doesn’t have, and it can give a different view of the same resume on a second run.
• It can bring the rules in: used to decide or rank who moves forward for a New York City job, it can count as an automated employment decision tool, with the audit and notice that come with one. In California, a covered business that uses its output to make hiring decisions, without a reviewer who weighs it and can change the result, falls under the privacy agency’s 2027 rules.
• The data leaves your systems: a resume is personal information. Check the account’s data terms before you paste one in, and use an account your agency controls.

The safer use is a summary for the recruiter, who reads the resume, makes the call and can explain it to the candidate and the client.

What is the best AI for recruiting?

There isn’t a single best tool, because recruiting is several jobs. Choose by the job:

• Your applicant tracking system’s own AI: search, matching and drafting inside the system your recruiters already use, where the candidate data already lives.
• A chat assistant on an agency account: first drafts of posts and outreach, and summaries of call notes, under data terms your agency controls.
• A scheduling tool: interviews booked from shared calendars.
• A candidate chatbot: answers to questions about open jobs, and interviews booked. A good one says it is automated, answers only from your own job and policy details, and hands off to a person on request.
• Workflow automation or custom software: for work that crosses systems your tools don’t connect, such as job orders that arrive by email and need to reach the tracking system.

Before you buy anything that scores or ranks candidates, ask the vendor five questions. What does it read? What does it decide? Can a recruiter change the result? Where is candidate data stored, and for how long? And is there a bias audit you could publish for New York City candidates?

What do candidates think about AI in hiring?

Many are wary. In a Harris Poll survey of 2,037 US adults for the American Staffing Association in June 2023, 49% of employed job seekers said AI recruiting tools are more biased than human recruiters. In the same survey, 39% of current job seekers said they had used AI tools to help them apply for a job.

The survey is from 2023, so treat the numbers as a starting point. For an agency whose business is candidates, that distrust matters. Telling candidates where AI is used, and that a recruiter makes the decision, is a small step, and for screening tools used on New York City candidates the notice is already required.

Will AI replace recruiters?

The government’s projections point the other way. The Bureau of Labor Statistics, in a handbook page last updated in October 2026, projects 6% growth from 2025 to 2035 for human resources specialists, the occupation that includes recruiters. That is 60,100 more jobs, with 73,700 openings a year on average. Its outlook for the occupation doesn’t mention AI.

What AI changes is the mix of the work: drafting, searching and scheduling can go faster, while the conversations with candidates and clients, and the judgment about fit, stay with people. The rules also look at the person in the decision: California’s 2027 privacy rules apply when no reviewer with the authority to change the decision analyzes the output, and New York City’s apply when a tool substantially assists or replaces the decision.

Where should a staffing agency start?

Start with a task that saves recruiter hours without judging anyone:

1. List the admin: for two weeks, note how recruiters spend their hours: posting, searching, scheduling, chasing paperwork and updating records.
2. Pick one task that isn’t a decision: interview scheduling, job order intake or record cleanup, because no candidate is judged by any of them.
3. Check your data: duplicate records and missing fields in your tracking system get copied by any tool you add.
4. Write down where AI may be used: which tools, for which tasks, on which data, and that a recruiter makes every hiring decision.
5. Ask before you buy screening: if a tool scores or ranks candidates, get its bias audit, check the notice rules where you recruit and talk to your employment lawyer.
6. Measure the hours: compare recruiter time on the task before and after, and keep the tool only if the hours come back.

To check your records and systems before you add a tool, use our AI readiness checklist. When the task is moving records between systems, the Zapier, Make, n8n or custom code comparison helps.

For help with the admin around placements, our page for staffing agencies says what we would build. The choice of who moves forward stays with your recruiters: we don’t build AI that screens candidates or decides who is hired. Talking it through starts with a 15-minute call.