Every resume that lands in a recruiter’s inbox used to be, at minimum, a person’s own attempt to describe their experience. That is no longer a safe assumption. Candidates can now use AI to write, polish and tailor applications at scale, while recruiters are increasingly using AI to screen them.
In the UK, the ICO has specifically warned that automated recruitment tools can create risks around fairness, transparency and personal data.
If AI is already helping you screen CVs, you should be able to see why a candidate matches, not just get a number.
EdgeTal uses semantic candidate matching to help recruiters find relevant candidates and see what contributed to each match.
The Resume Builder Problem
There is nothing wrong with using a resume builder or resume template to present genuine experience more clearly. The problem is that AI-assisted writing can make almost any CV sound polished, confident and closely matched to a job advert.
That makes surface-level quality less useful as a screening signal. A resume scanner or ATS resume checker can identify relevant wording without necessarily understanding whether the candidate has the depth of experience behind it.
Beating the System Is No Longer Hypothetical
“Beat applicant tracking system” is a real search because getting through automated screening is a real concern for jobseekers. Candidates can use keyword-stuffing and other tactics to make an application appear more relevant to an automated system.
More concerning is prompt injection. Researchers studying automated resume screening have found hidden instructions inside real resumes designed to influence the AI assessing them. One 2026 study analysed around 200,000 resumes and found hidden prompt injections in approximately 1% of them.
A separate 2026 study also found that prompt injection could affect applicant rankings in controlled AI screening environments, although its effectiveness varied depending on the screening setup and number of manipulated applications.
Where This Leaves Recruiters Using AI
If an AI written CV sits next to a genuinely human-written one and both are assessed by the same system, the important question is not whether AI was involved. It is whether the screening process can still assess the candidate on relevant evidence.
This also connects to AI bias in hiring. If a system relies heavily on patterns associated with previous hiring decisions, polished wording or other superficial signals can become part of what gets rewarded. The ICO has highlighted the need for organisations using automated recruitment to consider fairness, transparency and the potential for people to be unfairly affected.
What Actually Helps?
AI screening does not need to be abandoned. Recruiters still need ways to handle large volumes of CVs. The better approach is to understand what the software is actually measuring and keep enough visibility to question the result.
Look for:
- Meaning-based matching: Semantic matching can assess the context behind experience rather than relying only on exact keywords.
- Clear match reasoning: A score is more useful when you can see what information produced it.
- Human review: Suspicious or unusual results should be easy for a recruiter to investigate rather than automatically accepting or rejecting them.
- Honest limitations: Resume manipulation and prompt injection are still active areas of research. No screening tool should imply that the problem has been completely solved.
Where EdgeTal Fits
EdgeTal's candidate search runs entirely on-device, using a private LLM rather than a cloud-hosted model. It matches candidates by semantic meaning rather than relying only on exact keywords.
Each result also shows a match percentage and the specific part of the candidate profile that contributed to the match. That gives recruiters something to check rather than asking them to blindly trust a number.
It is important to be clear about the limits. EdgeTal does not claim to reliably detect every AI-written resume or hidden prompt injection. Those remain difficult problems across AI-based recruitment systems.
Final Thoughts
Your resume screening process is only as useful as the information it gives you to work with. AI can help recruiters search through CVs faster, but a polished resume, keyword-heavy application or high screening score should not become a substitute for checking the evidence behind the candidate. The goal is not to know whether AI touched the CV. It is to make sure your recruitment process can still see what actually matters.
FAQs
Can AI tell if a resume was written by AI?
Not reliably. AI-generated text can be difficult to distinguish from human writing, so AI-written resume detection should not be treated as definitive evidence.
Can candidates manipulate AI resume screening?
Yes. Research published in 2026 has documented prompt-injection techniques in resumes that attempt to influence AI screening systems.
Is an ATS resume checker enough for recruitment screening?
Not on its own. An ATS resume checker can identify keywords, qualifications and other criteria, but recruiters still need visibility into why a candidate matched and whether the evidence supports the result.
What should recruiters look for in AI recruitment software?
Look for transparent matching, meaningful candidate search, human review options and clear information about how candidate data is processed.