Most organizations do not have an AI candidate problem. They have an AI policy lag.
LinkedIn's 2026 research says 81% of people have used or plan to use AI in their job search. SHRM reports recruiting executives expect candidate AI use in applications and interviews to increase. Recruiters will keep encountering AI-assisted materials whether the employer has decided how it feels about them or not.
Not all AI assistance is the same.
Using AI to tighten résumé language is different from inventing experience. Practicing interview questions is different from receiving real-time hidden answers during an assessment. Using a search agent to discover roles is different from authorizing it to make representations the candidate never reviewed.
Policies should distinguish assistance from deception.
“No AI” is often too vague to enforce fairly.
Does spell-check count? A résumé rewrite? Translation support? Interview practice? A generated cover letter? A coding assistant during a take-home test?
If the employer cares about a boundary, define the behavior, not just the technology.
Tell candidates what matters before the test begins.
If AI is prohibited during a specific assessment, say so. If AI-assisted writing is acceptable but candidates must be able to defend the content, say so. If the employer uses AI in screening, explain the process where appropriate.
Ambiguity invites inconsistent enforcement and erodes trust on both sides.
Design assessment around capability.
The strongest answer to polished AI-generated material is not a detector. It is job-relevant evidence the candidate can explain, demonstrate or apply.
If the role requires judgment, give the person a scenario and ask how they would reason through it. If it requires writing, ask them to discuss choices in the work. If it requires technical execution, create a realistic task with clear rules about what tools are permitted.
Recruiters need a policy they can actually use.
Without guidance, each recruiter becomes a policy maker. One may reject anything that feels AI-written. Another may ignore obvious misuse. A third may be unsure whether a candidate should be confronted.
Create a simple matrix: acceptable assistance, restricted use, prohibited behavior and escalation required.
Do not mistake polish for dishonesty.
AI has made application materials more polished. That does not mean the candidate lacks the underlying capability. It means employers need stronger ways to validate the capability than relying on whether a résumé “sounds authentic.”
Do not mistake disclosure for complete protection either.
A policy alone will not solve automated mass apply, hidden interview assistance or fabricated experience. Controls, job-relevant assessment and human judgment still matter.
What should change this week?
Take the five most common ways candidates use AI in your process and decide what the employer actually cares about in each case. Write the rule in language a candidate and recruiter would interpret the same way.
Next step: Read AI Can Improve Hiring Quality. It Can Also Industrialize a Bad Decision. and The Candidate Is Real. The Application Still May Not Mean What You Think.