A surge in polished, low-effort, or automated applications can make an already demanding search feel unmanageable. Yet the practical challenge is not simply to remove anything that appears AI-assisted. Hiring teams need a process that separates application quality, eligibility, and evidence of job-related capability without treating legitimate candidates as suspect by default.
Distinguish automation from assistance and fraud
These categories should not be collapsed. A generic cover letter may reflect weak effort, a reusable template, an inexperienced applicant, or a candidate using generative AI to organize prose. None of those facts alone establishes deception. Conversely, fabricated credentials, impersonation, or coordinated automated submissions raise different concerns and warrant a different response.
Application volume is a real operational issue, but prevalence claims need care. In an August 2026 Breezy report, the vendor said 230,874 of 3,261,897 applications in its December 2025–May 2026 sample were flagged as bot-automated, or 7.08%. That is roughly one in fourteen applications in that vendor’s sample. It is not a verification that each flagged application was fraudulent, nor a general estimate for business school hiring, all industries, or every applicant tracking system.
Set defensible requirements before screening
The best protection against both bot volume and arbitrary screening is a clear definition of what the role actually requires. Identify a limited set of minimum qualifications, essential competencies, and disqualifying conditions that are genuinely connected to successful performance. For an academic program coordinator, that may include calendar management, stakeholder communication, and reliable handling of confidential information. For a faculty role, it may include scholarly expertise, teaching capability, and the credentials required by the institution.
Avoid turning preferred signals into hidden minimums. Rigid keyword rules can exclude candidates who describe comparable experience differently, especially across sectors, countries, or career paths. Where appropriate, allow equivalent experience and make that flexibility explicit in the review guidance.
Applicant tracking systems can support consistent administration, but their configuration matters. Periodically audit a sample of rejected applications, particularly those rejected for missing keywords or formatting issues. Review whether strong candidates were screened out and whether the same types of false positives recur. The exercise can reveal an overly narrow question, an ambiguous requirement, or a workflow that treats incomplete data as a negative decision.
Use suspicious signals as review triggers, not verdicts
Unusual submission rates, highly repeated text, inconsistent application details, or suspicious network activity may justify a closer look. They should trigger proportionate human review rather than automatic rejection. Shared IP addresses, VPN use, templates, recently created résumés, and polished language all have legitimate explanations.
Create an escalation path for applications that merit review. A trained recruiter can check for internal inconsistencies, compare qualifications with the posting, and request clarification when needed. The record should explain the job-related reason for the action taken, not label an applicant as dishonest based on a technical indicator alone.
Any technical controls should use approved tools and the minimum data necessary. Involve privacy, information security, and legal stakeholders where required by institutional policy. Do not rely on covert tracking, and do not upload résumés or applicant materials to unapproved AI services merely to assess whether text “sounds” machine-generated.
Collect better evidence earlier in the process
When volume is high, a short, role-specific assessment can be more useful than another round of résumé filtering. The assessment should be consistent, time-bounded, and clearly connected to the job. It should test capability, not polish alone.
Role area | Appropriate short assessment |
|---|---|
Admissions or student services | Draft a response to a realistic prospective-student question using provided information. |
Operations or program management | Prioritize scheduling conflicts and explain the proposed communication plan. |
Faculty or instructional roles | Lead a brief discussion of a teaching scenario or analyze a realistic case. |
Use a structured scoring rubric with defined criteria such as accuracy, prioritization, clarity, and job-relevant judgment. This is more defensible than broad impressions about confidence, conversational style, or whether someone appears “authentic.” Do not infer honesty, competence, or protected traits from a candidate’s face, voice, accent, appearance, or comfort on video.
State whether AI tools are permitted, limited, or prohibited for each exercise, and explain why. Offer accessible equivalent formats and accommodations. A short written response, phone conversation, or other accessible option may be appropriate when a video exercise would add little job-related evidence. Avoid requesting unpaid work that produces usable institutional output.
Keep people accountable for decisions
Automation can assist with administrative tasks such as organizing applications, sending acknowledgments, scheduling, and routing materials against defined requirements. It should not replace recruiter and committee judgment about qualitative evidence or final selection decisions. AI tools do not eliminate bias, and no detection method can reliably identify every automated application.
Human reviewers should receive concise guidance on how to apply criteria consistently, document exceptions, and avoid letting a polished narrative outweigh missing essentials. Just as important, candidates need a clear route to correct an application error, clarify a credential, or request an accommodation. That safeguard can prevent a data-entry mistake or automated flag from ending consideration prematurely.
Implementation checklist
- Define essential qualifications and acceptable equivalent experience before opening the search.
- Configure screening rules around those requirements, then audit samples of rejected applications for false positives.
- Use technical anomalies to prioritize human review, not as standalone grounds for rejection.
- Add one brief, structured, job-related exercise or conversation when it will improve the evidence available.
- Publish expectations for AI use, time limits, accommodations, and candidate correction requests.
- Limit data collection, use approved systems, and retain human responsibility for evaluation and decisions.
Monitor the process, not just the inbox
A workable response improves over time. Track application-to-completion rates, the number and outcome of escalated reviews, false positives found in audits, candidate feedback, accommodation requests, and the consistency of reviewer scores. Look for signs that controls are deterring genuine applicants or creating burdens that are not producing better hiring evidence.