AI is the flood.AI is the filter.
We asked 150+ talent acquisition leaders at companies with 5,000+ employees what AI is doing to their applicant pools and how they're responding. Applications are up, resumes are harder to trust, and budgets are moving toward screening and fraud checks.
The TLDR
are confident they catch bad applicants early
Yet 73% of them still lose 6+ hours a week.
01 The flood
Volume is up. So is misrepresentation.
Seven in ten TA leaders (71%) saw application volume per open role rise over the past 12 months. 23% say it rose significantly or more than doubled. Only 11% saw it fall.
Much of that volume is machine-assisted. 74% believe more than a quarter of the applications they receive contain AI-generated or AI-assisted content. 28% believe more than half do.
The sharper problem is accuracy, not volume. More leaders report a rise in candidates overstating skills or experience (77%) than a rise in applications per role (59%). Half (52%) now see candidates using AI live during interviews, and 40% report more fraudulent or identity-fraud applicants.
Change in application volume per open role
Past 12 months
Q6.
Share of applications believed to contain AI-generated or AI-assisted content
Percent of respondents choosing each range
Q7.
What increased in the past 12 months
Select all that apply
Q9.
02 The cost
About nine hours a week, per recruiter.
Asked how many hours a typical recruiter spends each week screening applicants who turn out to be unqualified, duplicate or misrepresenting themselves, the most common answer is 6–10 hours (46%). 70% say it's 6 hours or more. 24% say more than 10.
Taking the midpoint of each range, the average is about 9 hours a week, close to a quarter of a 40-hour week. That time goes to candidates who were never going to be hired. Leaders who believe more than half of applications are AI-assisted estimate 11.3 hours; everyone else, 8.1.
The confidence paradox
85% are somewhat or completely confident their process reliably identifies which candidates are real and qualified before recruiter time is spent. The hours say otherwise: 73% of confident leaders report 6+ hours lost per recruiter each week, and the completely confident report the most time lost of any group (9.7 hours).
Confidence in the process is not the same as efficiency in the process.
Hours a recruiter spends each week on applicants who turn out unqualified, duplicate or misrepresented
Percent of respondents choosing each range
Q8. Midpoints: under 2 = 1, 2–5 = 3.5, 6–10 = 8, 11–20 = 15.5, 20+ = 25.
Confidence that the process identifies real, qualified candidates before recruiter time is spent
Q11.
Hours lost per recruiter each week, by confidence level
Average of range midpoints
Q8 by Q11.
Cost calculator
What are bad applicants costing your team?
TA leaders estimate a recruiter loses about 9 hours a week screening applicants who turn out unqualified, duplicate or misrepresented. Enter your own numbers to see what that adds up to over a year.
Your estimate
of recruiter time spent on applicants who never should have reached them
Estimate. A full-time recruiter is counted as 40 hours a week for the working weeks entered. The survey estimate averages the midpoint of each answer range (Q8).
03 The filter
AI is in the stack. The waste is still in the funnel.
97% of TA teams use AI somewhere in hiring, in 4.3 parts of the process on average. Resume screening and ranking leads (73%), followed by interview scheduling (61%), identity or credential verification (54%) and interview intelligence (53%).
Adoption is heaviest after the application arrives. Programmatic media buying, the AI that shapes who sees a job in the first place, is the least-used application of AI in the survey (19%).
Heavier adoption hasn't bought back recruiter time yet. Teams using AI in four or more areas estimate 10.4 hours lost per recruiter each week; teams using it in three or fewer estimate 7.0. One likely reading: the teams under the most pressure adopted fastest.
Asked to name the AI tools they use in hiring, half (49%) name a general-purpose assistant such as ChatGPT, Gemini, Claude or Copilot. 13% name a purpose-built recruiting AI product.
Where teams use AI today
Select all that apply
Q10.
Hours lost per recruiter each week, by how widely AI is used
Number of hiring areas using AI; average of range midpoints
Q8 by Q10.
AI tools respondents named, by type
Open-ended answers, coded by the type of tool mentioned
Q24.
04 Trust
Trust moved from the page to the proof.
Asked which hiring signals they trust more or less than two years ago, leaders marked down the written application. The resume splits three ways: 33% trust it more, 36% the same, 31% less. The cover letter is the only signal to lose trust on net (−8).
The biggest gains went to signals that are witnessed, tested or confirmed by someone other than the candidate. Live interviews gained the most (+55), followed by internal referrals (+47), verified employment or credential data (+43), assessments (+36) and work samples (+35). Signals the candidate controls gained least: recorded video interviews (+18) and references (+16) trail far behind their live and verified equivalents.
Change in trust by hiring signal, past two years
Net trust = percent trusting a signal more minus percent trusting it less
Q12–Q20.
05 Spend
The next dollar buys screening and proof.
96% plan to increase TA spend somewhere over the next 12 months, across 3.4 areas on average. AI screening and matching leads (70%). Identity and fraud verification is second (53%), ahead of job advertising (50%), assessments (46%) and the career site and apply experience (45%). 40% plan to add recruiter headcount.
Fraud drives the verification budget. Among leaders who saw fraudulent or identity-fraud applicants rise, 69% will increase spend on identity and fraud verification, versus 43% of those who didn't.
Where spend will increase over the next 12 months
Select all that apply
Q21.
Plan to increase identity and fraud verification spend
By whether fraudulent or identity-fraud applicants rose
Q21 by Q9.
06 Agents
AI agents are starting to apply. Most leaders will let them in.
As AI agents begin applying to jobs on candidates' behalf, 70% of leaders would let them in on some terms, and 27% would block them. 35% would allow but flag them, 18% would treat them the same as human applicants and 18% would build an experience specifically for them. Only 3% haven't considered it.
Scale breeds caution. At companies with 10,000+ employees, 32% would block agents and 9% would build for them. At 5,000–9,999 employees, it's 22% and 28%.
Experience with fraud hardens the line. 36% of leaders who saw fraudulent applicants rise would block agents, versus 21% of those who didn't.
How organizations should respond to AI agents applying on candidates' behalf
Ordered from closed to open
Q22.
07 In their words
What they want next: AI that can tell who's real.
Asked in their own words how AI has changed their recruiting, half (49%) describe gains: faster screening, less admin, better shortlists. A quarter (25%) describe costs: more volume and AI-made applications, more doubt about who is real or qualified, and more time spent vetting.
Asked what AI capability they wish they had, the most common request is detection. 43% of those who named a capability want help spotting fraud, fakes and AI-generated applications, or verifying identity, credentials and skills.
Open-ended answers were theme-coded by hand across all 159 responses; one answer can carry more than one theme. Quotes are a selection, edited for spelling and grammar.
Q23.
08 What it means
Five moves for TA leaders.
Verify before you screen.
Move identity and credential checks to the front of the funnel. 70% of leaders say recruiters lose 6+ hours a week to applicants who never should have reached them.
Measure the waste, not the confidence.
Track recruiter hours spent on applicants who wash out, plus duplicate and fraud rates per requisition. 85% are confident in their process; the hours don't reflect it.
Rebalance the signal stack.
Weight what is witnessed, tested and verified above what is written. Build earlier moments where candidates show their work rather than describe it.
Shape the pool before it arrives.
Most AI investment filters applications after they land. Only 19% use AI in programmatic media buying, the lever that decides who applies in the first place.
Write the agent policy now.
Leaders are split on agent applicants. Decide how agent-submitted applications are identified, flagged and routed before volume makes the decision for you.
About the survey
Online survey of 159 talent acquisition decision-makers, fielded September 2026. All respondents work at organizations with 5,000+ employees, hold Director-level roles or above, and are responsible for or involved in talent acquisition.
Hours estimates use range midpoints, with 20+ counted as 25. Open-ended answers were theme-coded by hand.
Percentages are rounded and may not sum to 100.
How can Recruitics help you with the AI in Hiring Challenges?
Talk to our teamEmployees
Annual revenue
Seniority