Using AI in Recruitment: What Works and the Rules
Back to Blog

Using AI in Recruitment: What Works and the Rules

AI in recruitment is the use of software to help with hiring work such as sourcing, screening, scheduling and assessment. Most HR and talent acquisition teams now use it somewhere. The hard part is not buying a tool. It is knowing where AI helps, where it fails, and what you owe candidates and regulators when you use it.

This guide is for HR, talent acquisition, procurement, tech and AI leaders at Canadian and US companies. It covers how AI is used in hiring, what the data says, how candidates use AI too, which rules apply, how to choose a tool, and a checklist for using it responsibly.

Key takeaways

  • AI use in HR is growing fast. SHRM found 43% of organizations used AI for HR tasks in 2025, up from 26% in 2024.
  • Candidates use AI as well. Gartner found 39% used it in their applications and predicts one in four candidate profiles worldwide will be fake by 2028.
  • Rules now apply. Ontario requires AI disclosure in public job postings for employers with 25 or more employees. New York City requires bias audits. The EU treats recruitment AI as high risk.
  • AI should support a human decision, not replace it. A resume screen is not identity, skills or background verification.

What is AI in recruitment?

AI in recruitment means using machine learning and generative AI tools to support the hiring process. These tools fall into three broad groups:

  • Matching and ranking tools compare candidate profiles with a role and sort them.
  • Generative AI tools draft job descriptions, outreach messages and interview summaries.
  • Automation tools book interviews and send status updates.

Teams searching for "AI for HR" or "AI recruiting" are usually looking for one of these. The right choice depends on which step of hiring is slowest or riskiest for you.

How is AI used across the hiring process?

AI can touch every stage of hiring. The table shows where it is common and what to check before you rely on it.

AI recruitment workflow stages from sourcing to hire
Where AI is used in recruitment and what to check
StageWhat AI doesWhat to check
Sourcing and matchingSearches databases and ranks candidate profiles against a roleCan the tool explain a ranking? Are the profiles verified?
Resume screeningReads resumes and scores themTest for hidden text and for uneven pass rates between groups
Scheduling and messagesBooks interviews and answers routine questionsKeep a person available for exceptions
Assessment and interviewsScores tests, runs structured AI interviews, transcribes and summarizesDisclose use. Check for bias and accessibility
DraftingWrites job descriptions and outreachEdit for accuracy. Remove inflated requirements

Sourcing and matching

Sourcing tools scan job boards, professional networks and talent databases to surface people you might not reach through postings. Good ones go past keywords and read skills in context, such as treating "Python development" and "Python programming" as the same skill. They can also surface passive candidates who are not applying anywhere. The weak point is the data behind the profile. A ranking built on unverified claims is only as reliable as the claims.

Screening

Screening tools read resumes, extract skills and rank applicants against a role. This saves the most time when volume is high. It carries the most risk too, because a ranking can quietly filter people out. Review the rankings for patterns, and keep a person responsible for who moves forward.

Scheduling and candidate messages

Chatbots and scheduling assistants answer routine questions, book interviews across calendars and send status updates at any hour. This is low-risk work that candidates tend to value, as long as they can reach a person when the bot cannot help.

How common is AI in recruitment?

It is common and rising. SHRM's 2025 Talent Trends research found that 43% of organizations used AI for HR tasks, up from 26% the year before. That figure covers HR work overall, not recruiting alone.

Share of organizations using AI for HR tasks Bar chart. 26 percent of organizations used AI for HR tasks in 2024. 43 percent did in 2025. 0% 25% 50% 26% 43% 2024 2025
Share of organizations using AI for HR tasks. Source: SHRM, 2025 Talent Trends (see source 1).

What does AI do well, and where does it fall short?

AI is strongest on high-volume, repeatable work. That includes first-pass sorting, scheduling, drafting and summarizing. These tasks eat recruiter time and carry low risk when a person reviews the output.

It is weaker where judgment and proof matter. A screening tool reads what a document says. It cannot confirm that the person did the work described, holds the skills listed, or is who they claim to be. Treat a high score as a lead, not a finding.

A practical split between AI and people
TaskBest handled byWhy
First-pass sorting of applicationsAI, with reviewSpeed and consistent criteria across large volumes
Scheduling and routine questionsAIAlways available, low risk
Confirming identity, skills, eligibility and backgroundVerification checks, with a person reviewing resultsA resume screen cannot prove any of these
Judging fit, context and potentialPeopleNeeds conversation and judgment
Final hiring decisionPeopleSomeone must be accountable for it

How are candidates using AI, and what does it mean for screening?

Candidates use AI too. In a Gartner survey of 3,290 candidates, 39% said they used AI during the application process, mostly to write resume text (54%) and cover letters (50%). Polished text is no longer a signal of effort or skill.

Gartner candidate survey findings on AI in hiring Horizontal bar chart. 52 percent of candidates believe AI screens their application. 39 percent used AI in their application. 26 percent trust AI to evaluate them fairly. 6 percent admitted to interview fraud. Believe AI screens their application 52% Used AI in their application 39% Trust AI to evaluate them fairly 26% Admitted to interview fraud 6% Share of surveyed candidates. Bars come from separate Gartner surveys (4Q24, 1Q25, 2Q25).
Candidate survey findings. Source: Gartner press release, July 31, 2025 (see source 2).

Hidden instructions in resumes

Some candidates add hidden text, often white text on a white background, that tells an AI tool to rate them highly. This is called prompt injection. In Greenhouse's 2025 survey, 41% of US job seekers said they had used it. Measured rates are lower: Greenhouse estimated about 1% of resumes it reviewed contained hidden text, while ManpowerGroup reported about 10% of the resumes it scanned (see source 3).

A simple check helps. Paste the resume text into a plain text editor and look for lines you could not see before. Better still, use a screening tool that strips or flags non-visible text before it scores anything.

Fake profiles and interview fraud

The bigger risk is identity. In the same Gartner work, 6% of 3,000 candidates admitted to interview fraud, meaning posing as someone else or having someone pose as them. Gartner predicts that by 2028, one in four candidate profiles worldwide will be fake. That is a forecast, not a current measurement.

Gartner recommends a layered approach. Set clear expectations about acceptable AI use. Use assessments that can surface fraud, such as in-person interviews. Extend checks beyond the first hiring step with identity verification and background checks. For a deeper look, read our guides on resume fraud detection and candidate identity verification.

How do candidates feel about AI in hiring?

Mixed, and often wary. Gartner found that only 26% of candidates trust AI to evaluate them fairly, and 25% said they trust employers less if AI is used to evaluate their information. In Greenhouse's 2025 survey, just 8% of job seekers said AI screening makes hiring fairer (see source 4).

Three things cause most of the friction. Candidates get no explanation for a rejection. They cannot reach a person when something goes wrong. And they do not know AI is involved at all. You can fix all three without slowing hiring down:

  • Say plainly where AI is used and that a person makes the final decision.
  • Give candidates a named contact for questions or accommodation requests.
  • Offer an alternative to an AI interview where a candidate asks, and tell them how long decisions usually take.

Does AI reduce bias in hiring?

It can go either way. A tool trained on past hiring decisions can repeat past patterns. In one widely reported case, Amazon built an experimental resume-ranking tool trained on a decade of mostly male resumes. Reuters reported that it penalized resumes containing the word "women's," and the project was scrapped. Amazon said the tool was never used by its recruiters to evaluate candidates (see source 5).

A well-designed and well-tested tool may apply the same criteria to everyone. The difference comes from testing, not from the label "AI."

A common first test compares selection rates across groups. Under the four-fifths rule of thumb used in US practice, a group's rate below 80% of the highest group's rate is a signal to investigate. It is a screen, not a verdict. Guidance from the US EEOC in 2023 discussed applying this rule to AI tools, so confirm current agency guidance with counsel (see source 10).

Litigation shows the stakes. In Mobley v. Workday, a US federal court conditionally certified a nationwide collective of applicants aged 40 and over in May 2025, letting age discrimination claims about an AI screening system move forward. Claims against the software vendor proceeded on an agent theory. These are allegations, and the case was still being litigated in the sources cited (see sources 8 and 9). The lesson for buyers is to ask vendors for evidence and not assume the risk sits with them.

What rules apply to AI in recruitment?

Selected rules on AI in hiring
WhereRuleWhat it asks of you
Ontario, CanadaEmployment Standards Act, in force January 1, 2026Employers with 25 or more employees must state in publicly advertised postings if AI is used to screen, assess or select applicants. Retain postings and prescribed application records for three years.
Canada, privacyPIPEDA and provincial laws such as Quebec's Law 25Candidate data is personal information. Collect, use and disclose it within these laws, including when an AI tool processes it.
New York CityLocal Law 144, enforced since July 5, 2023Use automated employment decision tools only if they had a bias audit in the past year. Publish a summary and notify candidates.
European UnionAI Act, Annex IIIRecruitment AI is high risk. The Digital Omnibus (Regulation (EU) 2026/1744) moved high-risk obligations from August 2, 2026 to December 2, 2027. Relevant if you hire in the EU.
United StatesExisting anti-discrimination law, tested in courtTools can draw claims under laws such as the Age Discrimination in Employment Act. See Mobley v. Workday above.

This is general information, not legal advice. Check your obligations with employment counsel.

How do you choose AI recruiting software?

Start with the problem, not the product. Name the one step that costs you the most time or carries the most risk, then look for a tool that fixes that step. Also decide what kind of tool you need. An applicant tracking system stores candidates and tracks their progress. An AI tool ranks, matches or screens them. Many products now do both, so ask which parts are AI and which are workflow.

AI candidate matching process

Use these questions when you compare vendors:

  • Explainability. Can the tool show why it ranked or rejected a candidate?
  • Bias evidence. Can the vendor share a recent bias audit summary and say who ran it?
  • Data sources. What does the tool learn from, and are candidate profiles verified?
  • Security and privacy. How is candidate data stored, how long is it kept, and can candidates ask for deletion?
  • Integration. Does it work with the systems you already use? The table below lists the usual connection points.
  • Human control. Can a recruiter override the tool and see an audit trail of what happened?
Integration points to check
IntegrationWhy it matters
Applicant tracking systemKeeps candidate data in one place. See our ATS integrations
Calendar syncLets interviews be booked without email back and forth
Communication toolsKeeps candidate messages consistent
Background check connectionsSpeeds up verification steps
Reporting and analyticsGives one view of outcomes, including fairness checks

Ask each vendor to run a demo on your own job descriptions and a sample of real, anonymized applications. Generic demos show features. Your data shows performance. Building your own tool can make sense for a large team with unusual requirements, data science staff and enough hiring volume. For most teams, buying and testing is faster and cheaper to maintain.

How do you roll out AI in recruitment and measure results?

Start small. Pick one high-impact use case, such as scheduling or first-pass screening. Run a pilot on a limited set of roles. Measure against your old process. Scale only what works.

  1. Assess bottlenecks. Use your own data to find where time and risk build up.
  2. Pick a use case. Match the tool to the bottleneck.
  3. Pilot with limited scope. Involve recruiters in setup and train them to question the tool's output.
  4. Measure. Record a baseline before launch.
  5. Scale and review. Expand slowly and review results on a schedule.
AI recruitment metrics dashboard
What to measure before and after
AreaExample measures
SpeedTime to hire, screening time per candidate, time to schedule an interview
QualityHiring manager satisfaction, first-year retention, performance ratings
CostCost per hire, recruiter hours per hire, agency fees
FairnessPass rates by group at each stage
Candidate experienceApplication completion rate, candidate satisfaction scores

We do not quote benchmark improvements here. Published vendor numbers vary widely and rarely match another team's roles, so your own baseline is the only fair comparison.

How do you use AI in recruitment responsibly?

Use this checklist before you add or expand an AI tool.

  1. Map where AI touches hiring. List each tool and the decision it influences.
  2. Keep a person accountable. A named person should own any decision that affects a candidate.
  3. Disclose use. Do it where the law requires it, and consider it where it builds trust. Ontario employers must say whether AI screens, assesses or selects applicants.
  4. Test for uneven outcomes. Compare pass rates across groups before launch and on a schedule after.
  5. Check inputs. Look for hidden text and formatting tricks before a tool scores a resume.
  6. Verify what a screen cannot. Confirm identity, skills, work eligibility and background separately.
  7. Protect candidate data. Collect only what you need, set a retention period and follow privacy law.
  8. Keep records. Save postings, application records and audit results.

Where does verification fit with AI recruiting tools?

AI can sort candidate profiles faster. It cannot confirm that the person behind a profile is real, has the skills listed, and is allowed to work where the role is based. Those are separate checks, and they matter most for senior AI, data and tech roles where one bad hire is costly. Our guide on how to stop consultant bait and switch covers one common failure when the person you interview is not the person who shows up.

At Augmnt, every consultant is checked for resume fraud, skills, work eligibility and background. We use proprietary tools to detect resume fabrication, bait-and-switch tactics and multiple job holding. Consultants are also scored on a 100-point system that covers technical skills, communication and project delivery. All talent is based in the US or Canada.

The Augmnt platform combines AI matching with ATS integrations and a dedicated Customer Success Manager. It gives businesses a shortlist of three to five verified candidate profiles per role. Our staff augmentation service sources, verifies and delivers onshore consultants in the US and Canada, with no platform subscription. To see how scoping and checkpoints work for AI roles, read How to Hire a Verified AI Consultant for Your Business, or request a demo.

Frequently asked questions

What is AI in recruitment?

AI in recruitment is the use of machine learning and generative AI to support hiring tasks such as sourcing, screening, scheduling, assessment and drafting. It supports people who make the final decision.

How common is AI in recruitment?

SHRM's 2025 Talent Trends research found 43% of organizations used AI for HR tasks, up from 26% in 2024. That covers HR overall, not recruiting alone.

Does AI reduce bias in hiring?

Not automatically. AI can repeat patterns in past data or apply one standard to everyone. Test selection rates across groups before launch and on a schedule, and keep a person accountable for decisions.

Do I have to tell candidates I use AI?

It depends on where you hire. Ontario employers with 25 or more employees must say in public job postings if AI screens, assesses or selects applicants. New York City requires candidate notice for covered tools. Confirm with employment counsel.

Can AI detect fake candidates?

Not on its own. Screening tools can flag patterns, but confirming identity, skills, work eligibility and background needs separate verification steps. Gartner recommends a layered approach.

What are prompt injections in resumes?

They are hidden instructions in a resume, often white text, meant to push an AI tool to rate a candidate highly. Pasting resume text into a plain text editor can reveal them, and some screening tools can strip them automatically.

How do I choose AI recruiting software?

Start with the step that costs you the most time or carries the most risk. Ask vendors for explainability, a recent bias audit summary, data and security details, and integration with your systems. Run a demo on your own job descriptions and pilot before you scale.

What should I measure after adopting AI in recruiting?

Record a baseline first, then track speed, quality, cost, fairness by group and candidate experience. Compare against your own earlier results, not vendor averages.

Sources

  1. SHRM. "2025 Talent Trends: AI in HR." 2025. shrm.org
  2. Gartner. "Gartner Survey Shows Just 26% of Job Applicants Trust AI Will Fairly Evaluate Them." Press release, July 31, 2025. gartner.com
  3. Built In. "Hidden AI Prompts in Resumes." October 15, 2025 (reports Greenhouse and ManpowerGroup figures). builtin.com. Survey figure: Fortune, "Hiring, job seekers and the AI doom loop," November 18, 2025, reporting Greenhouse's 2025 AI in Hiring Report. fortune.com
  4. Fortune. "Hiring, job seekers and the AI doom loop." November 18, 2025 (8% of job seekers say AI screening makes hiring fairer, per Greenhouse). fortune.com
  5. Reuters. Jeffrey Dastin. "Amazon scraps secret AI recruiting tool that showed bias against women." October 10, 2018. reuters.com
  6. Osler, Hoskin & Harcourt. "AI in hiring: Ontario employers grappling with new job posting disclosure requirement." February 3, 2026. osler.com
  7. Williams HR Law. "Ontario's New Job Posting and Hiring Transparency Rules in Force as of January 1, 2026." January 12, 2026 (record retention). mondaq.com
  8. Holland & Knight. "Federal Court Allows Collective Action Lawsuit Over Alleged AI Hiring Bias." May 27, 2025. hklaw.com
  9. Maynard Nexsen. "Emerging Liability for AI-Driven Hiring Tools: Key Developments in Mobley v. Workday, Inc." April 23, 2026. maynardnexsen.com
  10. McGuireWoods. "EEOC Issues Guidance on Artificial Intelligence Hiring Tools." 2023 (four-fifths rule). mcguirewoods.com
  11. Norton Rose Fulbright. "New York City begins enforcing restrictions on the use of artificial intelligence." 2023 (enforcement from July 5, 2023). nortonrosefulbright.com. See also Office of the New York State Comptroller, "Enforcement of Local Law 144," December 2, 2025. osc.ny.gov
  12. Mondaq. "The AI Act After the Digital Omnibus: Simplified Rules, Delayed Deadlines." July 2026 (Regulation (EU) 2026/1744). mondaq.com. See also Ogletree Deakins on the delay: ogletree.com
  13. Mondaq. "What Canadian Employers Need to Know About Artificial Intelligence in the Workplace by 2026" (PIPEDA and Quebec Law 25). mondaq.com

About the author. Keara Garrett writes about hiring, candidate verification and AI talent for Augmnt, a Toronto-based marketplace that connects businesses with verified AI, data and tech consultants. Last reviewed .