Home / AI for HR / AI recruitment tools
ai hiring tools
AI recruitment tools
AI recruitment tools are products that use machine learning or generative AI to source, screen, assess, schedule or communicate with job candidates.
This page sets out the categories and a checklist for evaluating them. It does not rank products. For HR tools beyond hiring, see AI tools for HR; for the stage-by-stage legal picture, see AI in recruitment; and for the wider people function, start from AI for HR.
A shortlist you can defend: each tool placed in a risk category, with evidence on bias, data protection, explainability and contract terms gathered before you sign.
ICO audits of AI recruitment tool providers made almost 300 recommendations; some tools allowed filtering by protected characteristics or inferred gender and ethnicity from names. ICO: intervention into AI recruitment tools ↗
Six questions before procuring: DPIA, lawful basis, controller/processor responsibilities, bias mitigation, transparency, limiting unnecessary processing. ICO: thinking of using AI to assist recruitment? ↗
New to the topic? Read the background on artificial intelligence.
ai hiring tools
what categories of AI recruitment tools are there?
Six broad types, from low-risk drafting and scheduling to high-risk scoring and video assessment.
Drafting tools write or edit job adverts and candidate messages. General assistants already in your workplace often do this well, and the risk is mainly biased wording, which a reviewer can catch.
Sourcing and advertising tools search for and target potential candidates. Screening and matching tools parse CVs and score or rank applications against a role.
Assessment tools run tests, games or recorded video interviews and analyse the results. Scheduling and chat assistants book interviews and answer candidate questions.
Many applicant tracking systems now include several of these features, so a single platform may contain both a harmless scheduler and a high-risk ranking model. Evaluate the features, not the brand.
Drafting and scheduling: lower risk.
Screening, ranking and assessment: higher risk.
Evaluate feature by feature.
ai hiring tools
how do we tier AI recruitment tools by risk?
By what the output does to a candidate: informing a person, shaping a shortlist or making a decision.
Tier one tools help a recruiter work faster but never touch who progresses, such as advert drafting and slot booking. Normal acceptable-use rules and spot checks are usually enough.
Tier two tools shape a shortlist that a person then reviews, such as ranking or summarising CVs. These need a DPIA, group-level outcome testing and a documented review step. The ICO expects a DPIA before using an AI tool, ideally at procurement.
Tier three tools make or effectively make the decision, such as automatic rejection. Under Article 22A to 22D of the UK GDPR, in force since 5 February 2026, these need specific safeguards: information, representations, human intervention and a route to contest.
Most employers should aim to keep AI recruitment tools in tiers one and two until they have the evidence and process maturity for tier three.
ai hiring tools
what bias testing evidence should a vendor provide?
Group-level results on data like yours, how they were measured and what the vendor did about any gaps.
Ask for evidence, not a statement. The ICO says employers should seek clear assurances that a provider has mitigated bias and ask to see relevant documentation.
Useful evidence shows selection or pass rates by group, the data the test used, when it was run, and what changed as a result. A single overall accuracy figure tells you nothing about fairness.
Ask how the tool obtains characteristic data for monitoring. The ICO's audits found tools inferring gender and ethnicity from names and told providers to collect accurate information directly from candidates instead.
Plan to repeat the test on your own vacancies during a pilot, because performance on a vendor's sample does not guarantee performance on your roles and applicant pool.
| Check | What to establish |
|---|---|
| 1 | Ask for group-level results. |
| 2 | Reject inferred characteristic data. |
| 3 | Retest on your own vacancies. |
ai hiring tools
what should the DPIA and data checks cover?
Purpose, lawful basis, data minimisation, retention, transfers and who is controller for each processing step.
The DPIA should be done before use and kept up to date. The ICO's six procurement questions are a practical frame: DPIA, lawful basis, controller and processor roles, bias, transparency and limiting unnecessary processing.
Check what data the tool collects beyond the application itself. The ICO found some tools building large candidate databases and keeping data indefinitely without candidates knowing.
Settle roles in writing. If a vendor uses your candidates' data to improve its own product, it may be acting as a controller for that purpose; the ICO's guidance on controllers and processors explains the test.
For the wider data protection approach to AI, including fairness and accuracy, the ICO's guidance on AI and data protection is the reference point.
ai hiring tools
how much explainability should AI recruitment tools offer?
Enough for a recruiter to explain an outcome to a candidate and to override it with a recorded reason.
If a candidate asks why they were rejected, someone needs to give an answer that goes beyond 'the system scored you low'. Ask the vendor what factors drive the output and how that is shown to users.
Explainability also supports meaningful human involvement. A reviewer who can see why a candidate was ranked low can disagree; one who sees only a number is likely to defer.
The government's Responsible AI in Recruitment guide treats this as part of assurance before and after deployment, alongside bias and accessibility checks.
Test explainability in the pilot by asking recruiters to explain five outcomes in writing. If they cannot, the tool is not ready for a stage that affects who progresses.
Ask what drives each score.
Make overrides easy and recorded.
Test explanations during the pilot.
ai hiring tools
what should vendor contracts for AI recruitment tools include?
Data roles, retention, model changes, testing access, audit support, accessibility and exit terms.
The contract should say who is controller and processor for each purpose, where data is stored, how long it is kept and whether it is used to train or improve the vendor's models.
It should require notice of material model changes, because a tool that passed your bias test can behave differently after an update. Ask for the right to re-test.
Include cooperation with candidate rights requests, regulator enquiries and your own audits, plus accessibility commitments and a supported adjustment route.
Finally, agree exit terms: return or deletion of candidate data and the format it comes back in. predeploy can help you prepare these questions, but legal review of the contract belongs with your lawyers.
Notice and re-test on model changes.
No training on your data without agreement.
Clear deletion on exit.
ai hiring tools
how do teams use AI recruitment tools well after purchase?
With training on the real process, monitoring by group and a named owner for each tool.
Name an owner for each tool who reviews outcome data each hiring round and can pause a feature that drifts.
Train the recruiters and hiring managers on what the outputs mean and when to override them. Our AI training for recruiters uses your vacancies and your systems, and AI training for HR covers the wider team.
Keep the candidate-facing information accurate as tools change, and keep a short record of why each automated step exists.
We do not sell, rank or certify AI recruitment tools, and no checklist guarantees compliance. The aim is evidence you can show a candidate, a tribunal or the ICO.
One named owner per tool.
Review outcomes every hiring round.
Keep candidate information current.
Make the decision concrete
What does the current task cost in time?
Choose a task connected to ai recruitment tools and enter its current workload. This gives you a baseline to discuss before buying training or commissioning a build.
A baseline is useful even if you decide to keep the process manual. Measure the work that matters, then compare quality and review effort as well as time.
Assess the workflow before changing it, or plan business AI training if the gap is how people use existing tools.
Your current workload
Estimate how much time the existing task takes before discussing ai recruitment tools.
16 hours per week
People × minutes × weekly tasks ÷ 60. A workload baseline from your inputs, not a prediction of AI time savings. Include review and support time when comparing a proposed workflow.
A useful AI investment leaves your team able to do the work and judge the result.
Continue with the right question
Where should you go next?
Use the overview for context, then follow the pages that match your delivery decision.
FAQ
Questions, answered
Evidence and further reading
Which sources support the guidance?
Source links support the specific background described here. The practical checklists and illustrative examples are predeploy’s guidance.
ICO audits of AI recruitment tool providers made almost 300 recommendations; some tools allowed filtering by protected characteristics or inferred gender and ethnicity from names. ICO: intervention into AI recruitment tools ↗
Six questions before procuring: DPIA, lawful basis, controller/processor responsibilities, bias mitigation, transparency, limiting unnecessary processing. ICO: thinking of using AI to assist recruitment? ↗
Published 25 March 2024; assurance good practice for procuring and deploying AI in HR and recruitment, covering sourcing, screening, interview and selection. DSIT: Responsible AI in recruitment guide ↗
Replaces UK GDPR Article 22 with Articles 22A-22D: solely automated means no meaningful human involvement; significant decisions need safeguards (information, representations, human intervention, contest); special category data restricted. Data (Use and Access) Act 2025, section 80 ↗
Defines controllers and processors under UK GDPR; a processor acts on the controller's behalf and can be an individual such as a consultant. ICO: What are controllers and processors? ↗
ICO guidance covers AI accountability, transparency and fairness; the guidance is currently under review. ICO: AI and data protection guidance ↗
Background reading on the broader field of artificial intelligence. Wikipedia: artificial intelligence ↗
Choose your next step
Explore ai for hr
Start with a topic overview, or go directly to the question your team needs to answer.
AI for HR
Recruitment, people work and the UK rules that apply, plus HR training.
Start with the decision
More questions in this topic
