Some ATS engines run AI detection on uploaded CVs in 2026. Most don't. The bigger risk isn't detection; it's that AI-drafted CVs read like AI-drafted CVs and get filtered by recruiters at the next step.
Workday added an 'AI authoring signal' to its parsing pipeline in late 2025. The signal is advisory; it doesn't auto-reject the CV, but it surfaces to the recruiter. Most other major ATS engines (Taleo, Greenhouse, Lever, iCIMS, Ashby) don't currently run an explicit AI detector on uploaded CVs.
AI text classifiers have a roughly 30% false positive rate on human-written text, which makes auto-rejection on the basis of detection untenable. Even Workday's signal is documented as advisory, not load-bearing.
Recruiters can spot AI-drafted CVs in 7 seconds, AI detector or not. The tells: vague verbs ('leveraged', 'spearheaded'), unquantified outcomes, identical sentence rhythms across every bullet, generic skills lists, summaries that sound like marketing copy.
The fix is to use AI for rewrites, not for drafting from scratch. Have the AI rewrite a bullet you've written but that's underperforming; don't have it write the bullet from nothing. The difference is night and day.
Most ATS engines can't and don't try. Workday added an AI authoring signal in late 2025 that's advisory only. The bigger risk isn't ATS detection; it's that recruiters spot AI-drafted CVs in 7 seconds and filter them out.
Yes, with a strong caveat. Use AI to rewrite specific bullets you've drafted yourself, not to write the whole CV from scratch. AI-rewritten bullets read like a sharper version of you. AI-drafted CVs read like AI-drafted CVs.
AI detectors have a roughly 30% false positive rate on human text. Auto-rejection on that basis is untenable, and no major ATS does it today. Some flag for recruiter review; the recruiter is the screen, not the detector.
Vague verbs ('leveraged', 'spearheaded', 'utilised'), unquantified outcomes, identical sentence rhythms, generic summaries that sound like marketing copy. Read every bullet aloud; if it sounds like a press release, rewrite in plainer language with specific numbers.
If asked, yes. Most hiring contexts in 2026 expect candidates to use AI as a writing aid. The concern is over-reliance, not use. Be transparent about the workflow if it comes up; the recruiter is screening for judgment, not for a refusal to use modern tools.
Free 60-second check. Deterministic, not an AI classifier.
Check my CV →