Home · Business · HR & People · Job ads & screening
As of 12 August 2026, AI cannot screen your job applicants.
This still needs a person who signs their name to it.
Can you do it?
5 minutesto a draft.
n/ait cannot be self-verified.
Cost, all in£0
Skill neededchat-fluent
Who has to check ita colleague
What the alternative costsA recruiter’s placement fee is commonly 15–20% of first-year salary; an experienced hiring manager screens a hundred CVs in a working day.
If this goes wrong: the model quietly filters out the candidates who would have been best, and the pattern in who it rejects becomes a discrimination claim with your company’s name on it.
What to actually do
Hand it to a person
The route this page recommends
A person who owns the outcome does this end to end, worth it when the failure is dear.
Use a tool built for this
Second choiceDo it yourself
The distant thirdA chat interface gets you a draft, but you cannot verify it yourself. That is the catch.
Prompt
Here is our job description and one CV with identifying details removed. List what in the CV matches each stated requirement and what is missing, citing the CV line for every claim. Do not rank candidates, do not score, and do not recommend a decision.
Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.
What it gets wrong
- Learns proxies for protected characteristics from its training data, so its rankings can discriminate without ever seeing a protected field.
- Its rejections are invisible: nobody interviews the strong candidate it screened out, so the worst errors never surface.
- Rewards CVs written the way models write, which selects for AI-polished applications rather than for ability.
What makes this a NO: legal accountability, judgement under ambiguity and stakes of error.
How we scored this
Five axes, each scored nought to two by hand: ten means AI carries the task cleanly, and the thresholds that turn a total into YES, PARTLY or NO are published in the methodology. Each axis name links to its definition.
| Axis | Score (0–2) |
|---|---|
| Output | 1 |
| Inputs | 1 |
| Verification | 0 |
| Liability | 0 |
| Effort delta | 1 |
| Total | 3 / 10 |
The methodology and its thresholds are published in full.
FAQ
- Can we use ChatGPT to screen CVs?
- No. The answer here is NO. The rejection decision carries legal weight under the Equality Act whoever computed it, the model’s errors are concentrated in the candidates you never meet, and no one can verify a shortlist without re-reading the pile it came from.
- Is AI CV screening legal in the UK?
- Software is not illegal; unaccountable decisions are the problem. Outcomes that disadvantage protected groups are discrimination under the Equality Act 2010 regardless of the tool. Solely automated rejections also engage UK GDPR rules on automated decision-making, which expect human review a rubber stamp does not satisfy.
- What can AI safely do in our hiring?
- The paperwork around the decision, never the decision. Requirement-matching summaries a human reads, job-ad drafting you edit, and scheduling. Every rejection is a person’s call, recorded as one.
- Why is AI bad at picking good candidates?
- It ranks by resemblance to past hires and polished writing, both of which are proxies. The candidates it wrongly rejects are invisible, so the error rate never shows up in any metric you watch. What it measures well is how much a CV reads like its own writing.
Nearby answers
The newsletter
AI news, new answers and product picks, straight to your inbox.