AI in recruitment means using software, including machine learning and generative AI, to assist or automate hiring tasks. That covers parsing CVs, sourcing and matching candidates, scheduling interviews, answering questions through chatbots, drafting job ads, and screening applicants at scale. The core idea is simple. AI assists the process, and humans should still make the final call.
This post gives you a balanced view. AI can save real time and cut some manual work. It can also repeat bias and create legal exposure if you let it decide on its own. Both things are true. Below you will find where AI genuinely helps, the documented risks, the laws that now apply, and a practical way to use AI responsibly.
Key takeaways
- AI is already common in hiring. 64% of organizations using AI for HR applied it to recruiting, interviewing and hiring (SHRM, 2024).
- The clearest wins are repetitive tasks: writing job descriptions, screening resumes, automating searches, and communicating with applicants (SHRM, 2025).
- AI can repeat bias from past data. Amazon scrapped a tool that penalized resumes with the word "women's" (Reuters, October 10, 2018).
- Several laws now apply: NYC Local Law 144, EEOC Title VII guidance, the Illinois AI Video Interview Act, and the EU AI Act.
- The EEOC settled its first AI hiring case against iTutorGroup for $365,000 over age-based auto-rejections (EEOC, September 11, 2023).
- Keep a human in the loop. Human oversight is required for high-risk recruitment AI under the EU AI Act.
Why it matters
AI is no longer a niche tool in hiring. In SHRM's 2024 Talent Trends research, recruiting, interviewing and hiring was the top HR area for AI use: 64% of organizations using AI for HR applied it there. In SHRM's 2025 research, just over half (51%) of organizations use AI to support recruiting. LinkedIn's 2025 recruiter survey found nearly two-thirds of talent teams using or testing AI in hiring. If you run an agency, your competitors and clients are already trying it.
The stakes are also rising on the legal side. Regulators in New York City, the wider United States, Illinois, and the European Union have all moved to set rules for hiring AI. Getting this right protects candidates from unfair outcomes and protects your business from penalties and lawsuits. Understanding both the upside and the limits helps you decide where AI fits and where a person must stay in charge.
Where AI genuinely helps
Parsing and extracting CV data
AI can read a resume and pull out structured data: name, roles, dates, skills and education. This turns messy documents into clean fields you can search and format. It is a low-risk assistive task because it organizes information rather than deciding who gets hired.
Drafting job descriptions and outreach
Writing job ads is the most common AI use in recruiting. In SHRM's 2025 research, 66% of organizations using AI for recruiting used it to write job descriptions, and 31% used it to customize job postings. A person should still review the wording before it goes live.
Sourcing and matching candidates
AI can search large pools and suggest candidates who match a role. SHRM's 2025 data shows 32% of organizations using AI for recruiting used it to automate candidate searches. Treat the results as suggestions to review, not a final shortlist.
Screening resumes at scale
AI can sort and surface applications faster than manual review. SHRM's 2025 research found 44% of organizations using AI for recruiting used it to screen resumes. This is also where bias risk is highest, so human review of the output matters most here.
Scheduling and answering questions
AI can handle scheduling and answer common candidate questions through chatbots. SHRM's 2025 data shows 29% of organizations using AI for recruiting used it to communicate with applicants. These tasks are mostly administrative and free up recruiters for relationship work.
More time for people work
LinkedIn describes generative AI as automating time-consuming tasks so recruiters can focus on relationships and candidate experience. The benefit is not replacing judgment. It is removing repetitive work so people can spend time where judgment matters.
The real risks and the legal landscape
Algorithmic bias (Amazon, 2018)
AI can learn and repeat bias in past hiring data. Amazon built an experimental recruiting engine starting in 2014 that rated candidates one to five stars. Trained on ten years of mostly male resumes, it taught itself to penalize resumes with the word "women's" and downgraded graduates of two all-women colleges. Amazon scrapped the tool by early 2017. Reuters reported it on October 10, 2018.
NYC Local Law 144 (effective 2023)
New York City requires that any Automated Employment Decision Tool (AEDT) pass an independent bias audit within the prior year, that the results be published publicly, and that candidates receive notice. The law was enacted December 11, 2021. The NYC Department of Consumer and Worker Protection began enforcement on July 5, 2023, with civil penalties of $500 to $1,500 per violation, per day.
EEOC Title VII guidance (May 2023)
On May 18, 2023, the EEOC released technical assistance on AI in employment selection. It states employers can be liable under Title VII if an AI tool causes adverse (disparate) impact, unless the tool is job-related and consistent with business necessity with no less-discriminatory, equally effective alternative. It applies the four-fifths rule as a "rule of thumb", not a final test.
EEOC settlement: iTutorGroup (September 2023)
The EEOC settled its first AI-related discrimination lawsuit against iTutorGroup for $365,000. The company had programmed its application software to auto-reject female applicants aged 55 or older and male applicants aged 60 or older, rejecting more than 200 qualified U.S. applicants because of age. The EEOC press release is dated September 11, 2023.
EU AI Act (high-risk recruitment)
The EU AI Act classifies AI for recruitment or selection as high-risk under Annex III, including tools that target job ads, filter applications, and evaluate candidates. It entered into force on August 1, 2024. Prohibited-practice and emotion-recognition rules began February 2, 2025. Core high-risk duties, including risk management, documentation and human oversight, apply from August 2, 2026.
Illinois AI Video Interview Act (2020)
Illinois Public Act 101-0260 took effect January 1, 2020. If you use AI to analyze video interviews for Illinois-based roles, you must notify applicants that AI may be used and explain how it works, get their consent before the interview, limit who sees the video, and destroy it and all copies within 30 days of a request.
How to use AI responsibly
Step 1: Map where you use AI and how much it decides
List each place AI touches hiring: parsing, sourcing, screening, scheduling, video analysis. Mark whether each one only assists or actually filters people out. Tools that substantially assist or replace human discretion may count as an AEDT under NYC Local Law 144 and as high-risk under the EU AI Act.
Step 2: Keep a human in the loop on decisions
A qualified person should review AI output, be able to override it, and stay accountable for the outcome. This human oversight catches errors, bias and edge cases. It is required for high-risk recruitment AI under the EU AI Act. The guiding rule is simple: AI assists, humans decide.
Step 3: Audit screening tools for bias
Run or request an independent bias audit on any tool that scores or filters candidates. Measure outcomes across protected groups using selection rates and the four-fifths rule. NYC Local Law 144 requires such an audit within one year before using an AEDT, with results published publicly.
Step 4: Give candidates notice and consent
Tell candidates when AI is used and, where required, explain what it evaluates and get consent. NYC Local Law 144 requires notice. The Illinois AI Video Interview Act requires notice, an explanation, and consent before AI-analyzed video interviews, plus deletion within 30 days of a request.
Step 5: Check the laws that apply to your roles
Your obligations depend on where candidates and roles are. Review NYC Local Law 144, the EEOC's Title VII guidance, the Illinois AI Video Interview Act, and the EU AI Act for any roles those cover. When unsure, get qualified legal advice for your jurisdictions.
Step 6: Document and review regularly
Keep records of which tools you use, how they were audited, and what notices you gave. The EU AI Act's high-risk duties include risk management and documentation from August 2, 2026. Good records also help you spot problems early and show you acted in good faith.
Do this
- Use AI for repetitive tasks like parsing, drafting and scheduling, and keep a person reviewing the output.
- Treat AI results as suggestions to check, not final decisions.
- Run an independent bias audit on any tool that scores or filters candidates.
- Tell candidates when AI is used, and get consent where the law requires it.
- Check which laws apply based on where your candidates and roles are located.
- Keep records of your tools, audits and notices.
- Make sure a qualified person can override any AI recommendation.
- Get qualified legal advice when you are unsure about your obligations.
Common mistakes to avoid
Letting AI auto-reject candidates
Allowing a tool to filter people out without human review is where the biggest risk sits. iTutorGroup's software auto-rejected applicants by age and led to a $365,000 EEOC settlement (September 11, 2023). Keep a person in the loop on every rejection that AI suggests.
Assuming AI is neutral
AI is only as fair as its training data. Amazon's tool learned to penalize the word "women's" because it was trained on mostly male resumes (Reuters, October 10, 2018). Garbage in, garbage out applies. Test outcomes across groups rather than trusting the tool by default.
Skipping the bias audit
Using an AEDT without an independent audit can break NYC Local Law 144, which carries penalties of $500 to $1,500 per violation per day since enforcement began July 5, 2023. Audit before you deploy, not after a complaint.
Forgetting notice and consent
Many rules require telling candidates AI is used. The Illinois AI Video Interview Act needs notice, consent, and deletion within 30 days of a request. Missing these steps creates exposure even if the tool itself works well.
Treating weak enforcement as a free pass
A December 2025 audit by the New York State Comptroller called Local Law 144 enforcement "ineffective". Weak enforcement is not a green light. The EEOC, the EU AI Act, and private lawsuits still apply, so build compliance in regardless.
Ignoring jurisdiction
Rules differ by location. A tool that is fine in one place may break the EU AI Act's high-risk duties or a U.S. state law elsewhere. Check the laws for each region where you hire before you roll a tool out.
Frequently asked questions
Is using AI in recruitment legal?
Yes, but with conditions that depend on where you hire. NYC Local Law 144 requires a bias audit, public results, and candidate notice for an AEDT. The EEOC can hold you liable under Title VII if an AI tool causes adverse impact. The EU AI Act treats recruitment AI as high-risk, with duties phasing in through August 2, 2026. Illinois has its own rules for AI video interviews.
What is an AEDT?
AEDT stands for Automated Employment Decision Tool, a term from NYC Local Law 144. It is any computational process from machine learning, statistical modeling, data analytics or AI that issues a simplified output, like a score or recommendation, used to substantially assist or replace human discretion in hiring or promotion. Under the law, these tools must pass an independent bias audit before use.
Can AI be biased in hiring?
Yes. Algorithmic bias happens when a tool produces systematically unfair outcomes against a protected group. It often comes from biased historical data. Amazon's experimental tool penalized resumes with the word "women's" because it was trained on mostly male resumes, and Amazon scrapped it. Testing outcomes across groups helps catch this.
What does "human in the loop" mean?
It means a qualified person reviews AI recommendations, can override them, and stays accountable for the decision, rather than letting the AI auto-decide. Regulators recommend it, and the EU AI Act requires human oversight for high-risk recruitment AI. The guiding rule is simple: AI assists, humans decide.
Does AI in hiring actually save time?
It can, for repetitive tasks. SHRM's 2025 research shows organizations use it to write job descriptions (66%), screen resumes (44%), automate searches (32%), and communicate with applicants (29%). LinkedIn describes it as automating time-consuming tasks so recruiters can focus on relationships. The benefit is removing manual work, not replacing judgment.
Is NYC Local Law 144 actually enforced?
It is on the books but weakly enforced in practice. A December 2025 audit by the New York State Comptroller, covering July 2023 through June 2025, called enforcement "ineffective", citing complaint-handling and inaccurate compliance reviews. Even so, the EEOC, the EU AI Act, and private lawsuits still apply, so it is not safe to ignore the rules.
The bottom line
AI in recruitment is neither a magic fix nor a trap to avoid. It genuinely helps with parsing, drafting, sourcing, scheduling and screening, and adoption is already high across the industry. At the same time, the Amazon and iTutorGroup cases show what happens when a tool decides on its own, and laws like NYC Local Law 144, the EEOC's Title VII guidance, the Illinois AI Video Interview Act, and the EU AI Act now set real expectations. The balanced path is to use AI for the repetitive work and keep a qualified person accountable for decisions.
A low-risk place to start is document parsing and formatting, an assistive task that organizes information rather than deciding who gets hired. RefineCV is one tool that does this. Wherever you apply AI, the same rule holds: AI assists, humans decide.
Start with the low-risk task
Parsing and formatting CVs organizes information, it does not decide who gets hired. RefineCV does that step, turning candidate CVs into a clean, branded layout. Try it free with 10 CVs, no credit card.
Related reading: CV parsing explained for recruiters and unconscious bias in hiring.
Sources
- SHRM, 2024 Talent Trends Survey: Artificial Intelligence Findings (2024): 64% of organizations using AI for HR applied it to recruiting, interviewing and hiring, the top HR area for AI use.
- SHRM, 2025 Talent Trends: The Role of AI in HR Continues to Expand (2025): Just over half of organizations (51%) use AI to support recruiting; common applications include writing job descriptions (66%), screening resumes (44%), automating candidate searches (32%), customizing job postings (31%), and communicating with applicants (29%).
- LinkedIn, The Future of Recruiting 2025 (2025): Nearly two-thirds of talent teams reported using or testing AI in hiring; LinkedIn frames generative AI as automating time-consuming tasks so recruiters can focus on relationships and candidate experience.
- Reuters / MIT Technology Review, Amazon scraps secret AI recruiting tool that showed bias against women (2018-10-10): Amazon's experimental AI recruiting engine (started 2014, rated candidates 1 to 5 stars) penalized resumes containing "women's" and downgraded two all-women colleges because it was trained on 10 years of mostly male resumes; scrapped by early 2017.
- NYC Department of Consumer and Worker Protection (DCWP), Automated Employment Decision Tools (2023-07-05): NYC Local Law 144 bars use of an AEDT without an independent bias audit within the prior year, public results, and candidate notice; enacted December 11, 2021, enforcement began July 5, 2023, with penalties of $500 to $1,500 per violation per day.
- EEOC Title VII technical assistance (summarized by Mayer Brown) (2023-05-18): On May 18, 2023, the EEOC released Title VII technical assistance stating employers can be liable for adverse impact from AI selection tools unless job-related and consistent with business necessity, applying the four-fifths rule as a rule of thumb.
- U.S. EEOC press release, iTutorGroup to Pay $365,000 to Settle EEOC Discriminatory Hiring Suit (2023-09-11): The EEOC settled its first AI discrimination lawsuit against iTutorGroup for $365,000; its software auto-rejected female applicants 55+ and male applicants 60+, rejecting more than 200 qualified U.S. applicants by age.
- EU Artificial Intelligence Act, Annex III (2024-08-01): The EU AI Act classifies recruitment and selection AI as high-risk under Annex III; it entered into force August 1, 2024, prohibited-practice and emotion-recognition rules began February 2, 2025, and core high-risk duties (risk management, documentation, human oversight) apply from August 2, 2026.
- Illinois General Assembly, Public Act 101-0260 (AI Video Interview Act) (2020-01-01): The Illinois AI Video Interview Act took effect January 1, 2020, requiring notice and explanation, applicant consent before the interview, limits on sharing, and destruction of the video within 30 days of request.
- Office of the New York State Comptroller, audit of Local Law 144 enforcement (2025-12-02): A December 2025 audit by the New York State Comptroller (covering July 2023 through June 2025) concluded enforcement of NYC Local Law 144 has been "ineffective", citing problematic complaint-handling and inaccurate compliance reviews.