AI for Recruitment in 2026: Where It Helps, Where It Fails, and the Rules
AI for recruitment means software that writes job posts, finds and screens candidates, books interviews and takes notes during them. It is now the most common use of AI in HR: SHRM’s 2026 research found that 27% of organizations use AI in talent acquisition, more than in any other HR function. The time savings are real for writing, scheduling and notes. Screening and ranking candidates is where the lawsuits are, and where the new rules in New York City, Illinois, Colorado and the EU apply.
Checked in October 2026. The legal summary below isn’t legal advice; check the rules with your own counsel. We make Realtimer, a tool for recruiters mentioned once below.
How companies use AI for recruitment today
SHRM surveyed 1,908 HR professionals for its 2026 State of AI in HR report. 39% of organizations use AI in HR, and 60% of those with more than 5,000 employees do. Recruiting leads every other HR function, and inside recruiting the top three uses are:
- writing and improving job postings
- screening and matching resumes
- scheduling interviews
LinkedIn’s Future of Recruiting 2025 report, as covered by HR Executive and Recruiting News Network, says recruiting teams using generative AI save about 20% of their work week, roughly one day. The same coverage says recruiters spend much of that time on candidate screening and skills assessment.
Company policies lag behind. SHRM found that fewer than half of organizations have a formal AI policy, and that 57% of HR professionals in states with AI workplace laws didn’t know those laws existed. About half of respondents saw no improvement in decision making from AI.
AI for recruitment, stage by stage
| Stage | What AI does | What to check |
|---|---|---|
| Job description | Drafts the post, rewrites it for clarity and tone | That requirements and salary are yours, not invented |
| Sourcing and outreach | Searches profiles, drafts personal messages | That messages mention something real about the person |
| Resume screening | Parses, matches and ranks applicants | Bias testing, and whether a recruiter can see why someone ranked low |
| Scheduling | Finds slots, books calls, sends reminders | Little; this is the safest place to start |
| Interview | Transcribes, writes notes, shows questions and the CV during the call | Consent to record, and no emotion analysis |
| Decision | Scores or summarizes candidates | That a person makes the decision and can explain it |
The risk grows as you go down the table. A bad job post costs you a rewrite. A bad ranking model can reject a whole group of qualified people, and you may not find out until someone sues.
Where AI for recruitment saves the most time
Writing is the easy win. A first draft of a job post, an outreach message or a rejection email takes seconds, and a recruiter who knows the role can fix it faster than writing from scratch. LinkedIn reports that companies using its AI-assisted messaging are 9% more likely to make a quality hire, a figure from LinkedIn about its own product.
Scheduling comes next. Coordinating three interviewers and a candidate across time zones is admin, and no hiring decision depends on it.
Then interview notes. When software writes the transcript and summary, the interviewer can listen instead of typing, and the hiring manager gets the same notes as everyone else. You still have to tell candidates the call is transcribed, and in some places get their consent.
Where it goes wrong
Screening models learn old bias
In 2018 Reuters reported that Amazon had scrapped an experimental resume-screening tool. It was trained on ten years of resumes sent to Amazon, most of them from men, and by 2015 the team found it wasn’t rating candidates for software developer and other technical jobs in a gender-neutral way. According to Reuters, recruiters looked at its recommendations but never relied on the rankings alone. Any model trained on your own past hires carries the same risk.
The vendor can be sued too
Mobley v. Workday is the case HR teams watch. Derek Mobley, a job seeker over 40, says Workday’s screening tools rejected him for more than 100 jobs. In May 2025 a federal judge in California let his age discrimination claim go forward as a nationwide collective action on behalf of older applicants. In March 2026 the court turned down Workday’s argument that applicants can’t bring disparate impact claims. As of mid-2026 there was no settlement.
The case targets the software vendor as well as the employer. If you buy a screening tool, ask how it was tested for bias and what the contract says about who is responsible.
The candidate may not be real
Gartner expects that by 2028 about one in four candidate profiles worldwide could be fake, according to HR Dive’s coverage of its July 2025 report. In a Gartner survey of 3,000 candidates, 6% admitted to interview fraud, either posing as someone else or having someone pose as them. In another survey of 3,290 candidates, 39% said they had used AI during the application process. We wrote a separate guide on spotting hidden AI help in interviews.
The rules in 2026
| Where | Rule | What it asks of employers |
|---|---|---|
| New York City | Local Law 144, enforced since July 2023 | An independent bias audit of the tool within a year before use, a public summary of the results, and notice to candidates |
| Illinois | Human Rights Act amendment (HB 3773), since January 1, 2026 | No AI that discriminates, no zip codes as a stand-in for protected groups, and notice when AI is used in hiring decisions |
| Colorado | AI Act, rewritten in May 2026 by SB 26-189 | Duties start January 1, 2027: notice, a way to contest an adverse decision with human review, and records kept for three years |
| European Union | AI Act | Emotion recognition at work banned since February 2025; high-risk duties for hiring AI moved to December 2, 2027 |
New York City
Local Law 144 covers automated employment decision tools. Employers and agencies must have the tool audited for bias no more than a year before using it, publish a summary of the audit and tell candidates the tool will be used. Penalties run from $500 to $1,500 a violation. In December 2025 the New York State Comptroller called the city’s enforcement “ineffective,” and law firms expect that criticism to push the city toward stricter enforcement.
Illinois
Since January 1, 2026, Illinois employers may not use AI that has a discriminatory effect in recruiting, hiring, promotion or discipline, and may not use zip codes as a proxy for a protected class. They must also tell employees and applicants when AI is used. The law leaves the details of that notice to the Illinois Department of Human Rights, which hadn’t finalized its rules when we checked.
Colorado
Colorado passed the first broad state AI law in 2024, then delayed it, and in May 2026 rewrote it with SB 26-189. The new version, effective January 1, 2027, asks much less of employers than the original. You tell people when AI is used in a consequential decision, give them a way to contest an adverse decision with a human review, and keep records for three years.
The EU AI Act
The AI Act treats AI used to recruit, screen or evaluate candidates as high risk. Those obligations were due on August 2, 2026, but the Digital Omnibus, Regulation (EU) 2026/1744, published on July 24, 2026, moved them to December 2, 2027. Two parts already apply. Since February 2, 2025, the Act has banned AI that infers emotions in the workplace, and the European Commission’s guidance counts job candidates as part of the workplace, so a video interview tool that reads a candidate’s face for enthusiasm or nerves is off the table in the EU. Since August 2, 2026, the transparency rules require that people know when they’re talking to an AI system, which covers recruiting chatbots.
Questions to ask before you buy an AI recruiting tool
- Does it decide, or does it suggest? If it rejects candidates on its own, you carry the legal risk under every rule above.
- Has it been audited for bias, by whom, and can you see the results? In New York City you need this anyway.
- Can a recruiter see why a candidate ranked where they did?
- What does it do with candidate data: where is it stored, for how long, and is it used to train the vendor’s models?
- What do candidates see? Ask for the notice text and the consent flow.
- Does it analyze faces, voices or emotions? In the EU, emotion recognition at work is banned.
- Can you test it on a past hiring round? Run it on the applicants for a role you already filled and compare its shortlist with the people you interviewed and hired.
AI in the interview itself
Most of the attention goes to screening, but a recruiter spends a lot of the week on calls. AI can help here without making any decision: a transcript, a summary for the hiring manager, the candidate’s CV and your question list on screen, and a check of the candidate’s answers against what they wrote in the application.
That’s the part Realtimer covers. It’s a real-time AI assistant for Mac that listens on the recruiter’s own computer, so no bot joins the interview, and it shows the CV, your question list and answers about the role from your files during the call, with notes afterwards. It doesn’t rank or score candidates; the decision stays with the recruiter. Details on the recruiting assistant page.
A safe order to adopt AI for recruitment
Start where a mistake is cheap: job posts, outreach drafts and scheduling. Add interview notes next, with a clear line in the invite that the call is transcribed. Leave automated screening and ranking for last, and only with a bias audit, a human who reviews every rejection, and a notice to candidates that matches the law where they live.
FAQ
What is AI for recruitment?
Software that helps with hiring tasks: writing job posts, sourcing and messaging candidates, screening resumes, scheduling interviews, taking interview notes and summarizing candidates for the hiring manager.
Is it legal to use AI to screen resumes?
Yes, but with conditions in some places. New York City requires a yearly bias audit and notice to candidates, Illinois bans AI that discriminates and requires notice, and the EU will treat screening AI as high risk from December 2, 2027. Anti-discrimination law applies everywhere, with or without AI.
Do I have to tell candidates I use AI?
In New York City and Illinois, yes, when AI is used in hiring decisions. Colorado will require notice from January 1, 2027. In the EU, people must be told when they’re talking to an AI system, such as a recruiting chatbot.
Will AI replace recruiters?
The tasks it handles well are writing, scheduling and notes. About half of HR professionals in SHRM’s 2026 research saw no improvement in decision making from AI, and the laws above push the hiring decision back to a person.
How do I check an AI recruiting tool for bias?
Ask the vendor for an independent bias audit and its results, then run the tool on a past hiring round and compare its shortlist with your own decisions, broken down by age, gender and ethnicity where you have that data.
Can AI analyze candidates’ emotions in video interviews?
Not in the EU. The AI Act has banned emotion recognition in the workplace since February 2, 2025, and the Commission’s guidance includes job candidates.
Sources: SHRM: AI in HR 2026; HR Executive: LinkedIn Future of Recruiting; LinkedIn: Future of Recruiting 2025; Reuters: Amazon recruiting tool; SHRM: Workday lawsuit; HR Dive: Gartner on fake candidates; NY State Comptroller: Local Law 144 audit; DLA Piper: NYC AI hiring law; Morgan Lewis: Illinois AI law; SHRM: Colorado amends its AI law; ActuIA: Omnibus 2026/1744; FPF: emotion recognition at work.