How to stop AI-generated and fake applications in Greenhouse

A setup order for Greenhouse admins: which native settings cut junk and fake applicants, what each plan includes, and where a human still has to decide.

The short answer

If you want to know how to stop AI-generated fake resumes in Greenhouse, start with the settings your plan already includes, then pay for more only where volume or risk calls for it. Turn up reCAPTCHA spam sensitivity on your job board, add a few required knockout questions, and auto-reject on the answers if you are on Plus or Pro. Cap applications per post and block repeat applicants. If Greenhouse spam applications still flood in, the Real Talent add-on adds a Spam Blocklist, a Fraud Detection report and identity verification with CLEAR. Keep in mind that the fraud tools flag, they don't reject, so a person still makes the call.

Why Greenhouse spam applications keep growing

A chatbot can tailor a resume in seconds, and bots can submit it at scale. In a November 2024 post, Greenhouse said the average job opening drew 222 applications in the first quarter of 2024, almost three times the level at the end of 2021.

The problem is not only volume. In a survey Greenhouse published in November 2025, 65% of hiring managers said they had caught applicants using AI deceptively, and 18% reported deepfakes. The same survey found that 34% of recruiters spend up to half their week filtering spam and junk applications. That is time not spent talking to real people.

No single setting fixes this. What works is layers: friction at the form, rules at intake, limits on repeats, and verification before interviews. Greenhouse has a native tool for most layers.

How to stop AI-generated fake resumes in Greenhouse: the setup order

Work through these in order. The early steps catch the most noise. The later ones cost more or add friction, so save them for the jobs that need them.

  1. Tune spam protection on each job board. Greenhouse uses Invisible reCAPTCHA on job boards and lets you set spam sensitivity per board. A stricter setting stops more bots but adds more friction for real applicants. The help article lists four careers-page integration options, so check which one your site uses.
  2. Add a few required custom application questions. Custom questions work on Core, Plus and Pro. Ask about work authorization, location or a must-have certification. Yes/No, single-select and multi-select questions are the useful ones, because rules can act on them.
  3. Turn those answers into application rules. On Plus and Pro, Auto-reject can reject an application based on a custom question answer, with a rejection reason, an optional email and a tag. If an application matches both an advance rule and a reject rule, the reject wins. On Core, use Auto-Tag to label the same answers so a recruiter can clear them in bulk.
  4. Cap volume per post. All three plans can set application capacity: a maximum number of applications per job post. Greenhouse can unpublish the post when it hits the cap and, if you choose, auto-reject the overflow with the reason "Rejected due to application capacity." You can also schedule a post to unpublish on a set date.
  5. Block repeat applicants (Pro). Application limits let you restrict repeat applications within a number of days you pick, and block people from reapplying after a rejection. A blocked application is auto-rejected and labeled as blocked by an auto reject rule.
  6. Decide on LinkedIn Easy Apply per job. LinkedIn Apply Connect is turned on job by job under Job Posts, then Publish to Free Job Boards. Greenhouse notes that since July 2025, free posts have an Easy Apply cap, after which applicants go to your careers site.
  7. Clean up duplicates and mark spam. Configure the Duplicate tag to match on name, email, phone or LinkedIn. Turn on auto-merge by email, with phone and LinkedIn as optional matches. When you see junk, use Reject as spam, which needs the Spam rejection reason to be visible in your settings.
  8. Add Real Talent where the problem persists. Real Talent is a paid add-on suite. Its Spam Blocklist auto-rejects at intake by email domain, email address or IP, before Fraud Detection runs, and admins manage it under Configure, then Real Talent, then Spam. Fraud Detection then builds a report from objective signals. Greenhouse treats a data-center IP as high risk and a time zone mismatch as a weak signal, and says the report does not use AI or automated scoring. It runs automatically on Pro and manually on Plus.
  9. Verify identity before the interviews that matter. Greenhouse identity verification with CLEAR, on Plus and Pro, asks the candidate for a government ID and a selfie liveness check inside MyGreenhouse. You can click Request verification on the Security screening tab, or have it sent automatically when a candidate reaches a stage you choose.
  10. Review what's left with a scorecard. Use interview kits and scorecards so reviewers judge the same criteria. Talent Matching, also part of Real Talent, scores applicants against a calibration in Application Review. Greenhouse says it does not automatically advance or reject anyone.

Feature, what it stops and which plan

Where Greenhouse's help article doesn't name a plan, we say so rather than guess. Confirm with your Greenhouse account team before you buy.

FeatureWhat it stopsPlan
Invisible reCAPTCHA and spam sensitivityAutomated bot submissions on your job boardPlan not named in the help article
Custom application questionsMass applicants who skip or fail must-have criteriaCore, Plus, Pro
Auto-TagJunk you can label and clear in bulkCore, Plus, Pro
Auto-reject rulesApplications with disqualifying answers, at intakePlus, Pro
Application capacity and scheduled unpublishUnbounded volume on a single postCore, Plus, Pro
Application limitsThe same person applying again and againPro only
Duplicate tag, auto-merge, Reject as spamDuplicate profiles and known spammers returningPlan not named in the help articles
Spam BlocklistKnown bad email domains, addresses and IPsReal Talent add-on
Fraud DetectionNothing on its own; it flags risk for reviewReal Talent; automatic on Pro, manual on Plus
Identity verification with CLEARPeople applying under someone else's identityPlus, Pro (part of Real Talent)
Talent MatchingReviewer time spent on poor matches; it ranks, not rejectsReal Talent add-on

The gaps to know before you rely on it

  • Flags are not rejections. Fraud Detection and Talent Matching inform a person. Neither one removes an applicant. Someone on your team still needs time to review the flagged list.
  • Auto-merge has blind spots. It needs an exact match, does not go back over old records, and skips applications that arrive through the API, from agencies, through referrals or on confidential jobs.
  • Reject as spam depends on auto-merge. A later application from the same email is auto-rejected only when it matches your auto-merge criteria. Anything that auto-merge skips can get through.
  • Friction has a cost. Stricter spam sensitivity, extra required questions and ID checks all add steps for honest applicants. Save the heavier controls for roles that attract fraud, such as remote technical jobs.

For remote technical roles, verification matters more than any spam filter. The FBI warned in 2022 about deepfakes and stolen personal data used to apply for remote IT jobs, and in January 2025 about North Korean IT workers using AI and face-swapping in video interviews.

Where Carolina fits after Greenhouse filters the junk

Once Greenhouse strips out the bots and repeats, you may still have hundreds of plausible applicants per job. That is where NTRVSTA works. Carolina, NTRVSTA's AI recruiting agent, connects to Greenhouse and scores every applicant against your scorecard, with written reasons for each candidate. She can then run a live first-round interview, by video or voice, at any hour. Her questions adapt to each answer, with follow-up questions on what the candidate just said. Scorecards and transcripts are written back to Greenhouse, where your team already works. See the Greenhouse integration page for how teams connect it.

Live follow-ups make memorized or relayed answers harder to keep up, and easier for a human reviewer to notice in the transcript. That is useful evidence, but it is not identity verification and it is not fraud detection. For identity, use CLEAR through Greenhouse, or your background check provider. Checkr, Sterling and HireRight all connect to NTRVSTA.

You stay in control of each step. Every action Carolina takes can be set to suggest, ask first or run on its own, and offers always wait for a person. For the wider playbook on junk applications, see our page on how to stop AI resume spam, or put numbers on your own funnel with the AI application spam cost calculator.

Where NTRVSTA isn't the right fit

If application capacity, auto-reject rules and Real Talent already keep each job's volume to something your recruiters can review in a day, adding another tool may not pay off. Greenhouse's own Talent Matching may be enough to sort what's left.

If your main risk is someone using a stolen identity, NTRVSTA is the wrong first purchase. Start with CLEAR verification and a background check, then see whether first-round screening is still your bottleneck. And if you mostly hire hourly staff through text-based apply flows, tools built for that model will suit you better.

Next, read our guides on proxy candidates in remote interviews and on spotting outside help during interviews. Running Lever instead? We wrote a separate guide to Lever spam applications.

Questions

Does Greenhouse automatically reject fraudulent applicants?

Not through Fraud Detection. Greenhouse says the Fraud Detection report flags applications using objective signals and does not reject anyone. Automatic rejection comes from other settings you choose: auto-reject rules on Plus and Pro, application limits on Pro, capacity overflow rejection, and the Spam Blocklist in Real Talent.

Which Greenhouse plan do I need to auto-reject spam applications?

Auto-reject application rules are available on Plus and Pro. Core includes Auto-Tag, custom application questions and application capacity, so a Core team can tag suspect answers and cap volume but cannot auto-reject on an answer. Application limits for repeat applicants are Pro only.

Why did a spam applicant get through after I marked them as spam?

Reject as spam only auto-rejects a later application from the same person when it matches your auto-merge criteria. Auto-merge needs an exact match, is not retroactive, and skips applications that come in through the API, agencies, referrals or confidential jobs. Those can still reach your pipeline.

Can NTRVSTA verify a candidate's identity in Greenhouse?

No. NTRVSTA does not verify identity or detect fraud. For identity, use Greenhouse identity verification with CLEAR on Plus or Pro, or your background check provider. NTRVSTA's agent, Carolina, scores applicants against your scorecard and runs live first-round interviews, and writes the results back to Greenhouse.

Is LinkedIn Easy Apply a big source of Greenhouse spam applications?

It can add a lot of low-effort volume. Greenhouse says that since July 2025, free LinkedIn job posts have an Easy Apply cap, after which applicants are redirected to your careers site. LinkedIn Apply Connect is enabled per job, so decide job by job whether you want it.

Want to see Carolina score a real Greenhouse pipeline against your scorecard? Book a demo and bring one noisy job. You can compare plans on our pricing page.

Sources

NTRVSTA
‍