Corporate AI8 min read

Can HR Use AI to Screen CVs in Kenya?

On this page

Yes, HR in Kenya can use AI to help screen CVs, but a person must make the decision. Kenya's Data Protection Act gives every applicant the right not to be subject to a decision based solely on automated processing that significantly affects them, and in our reading, being rejected for a job is that kind of decision. Use AI to organise and summarise applications against criteria you set, tell applicants you use it, test it for bias, and keep a named person deciding who goes through.

This guide is for HR managers, recruiters and business owners in Kenya. It is general information, not legal advice.

What the law says, in plain words

The rule on automated decisions

Section 35(1) of the Data Protection Act, 2019 says every data subject "has a right not to be subject to a decision based solely on automated processing, including profiling, which produces legal effects concerning or significantly affects the data subject".

There are three exceptions in section 35(2): where the decision is necessary for entering into or performing a contract with the person, where a law authorises it with suitable safeguards, or where it is based on the person's consent.

If an organisation does take a solely automated decision with significant effects, section 35(3) says it must tell the person in writing, as soon as reasonably practicable. The person can then ask it to reconsider, or to take a new decision that is not based solely on automated processing.

What the regulations add

Regulation 22 of the Data Protection (General) Regulations, 2021 describes automated individual decision making as a decision "made by automated means without any human involvement". Where it is used, the organisation must, among other things, inform the person, give "meaningful information about the logic involved", prevent errors, process data "in a way that eliminates discriminatory effects and bias", and make sure the person can obtain human intervention and give their point of view.

Regulation 49 lists automated decision making with legal or similar significant effects, and processing sensitive personal data, among the processing that is high risk. Under section 31 of the Act, high-risk processing needs a data protection impact assessment (DPIA) before it starts.

Discrimination law applies too

The Employment Act, 2007 bars direct or indirect discrimination against a "prospective employee", covers "selection criteria", and puts the burden on the employer to prove that discrimination did not take place. If a screening tool quietly filters out a group of people, expect the employer to answer for it, even if a vendor built the tool. Our HR training guide covers these sections and the ODPC's draft AI guidance in more depth.

What counts as "solely automated"?

Our reading, which you should check with your lawyer:

  • A tool rejects applicants and nobody looks at them. That is solely automated, even for a simple knockout rule such as "no degree, auto-reject". The Act speaks of automated processing, not only of AI.
  • A tool ranks 300 CVs and a recruiter only reads the top 20. The bottom 280 were in effect decided by the tool. Treat this as high risk.
  • A recruiter clicks "approve" on every AI suggestion without reading. Regulation 22 speaks of decisions "without any human involvement". A rubber stamp is thin involvement. Make the review real: the person reads the CV and can disagree.
  • AI summarises each CV against your criteria, and a recruiter reads every application and decides. This is the safer pattern.

Consent is one of the exceptions in section 35, so some employers ask applicants to "agree" to AI screening. Be careful. The Act defines consent as an "express, unequivocal, free, specific and informed" indication of a person's wishes. Section 32 puts the burden of proving consent on the organisation and lets people withdraw it at any time.

A job seeker told "agree, or we will not consider your application" may not feel free to say no. In our view, do not rely on a consent box to let a machine decide. Keep a person in the decision instead.

Tell applicants what you do

Section 26 gives people the right to be informed of how their data will be used. Section 29 says that before collecting data, an organisation must, as far as practicable, tell people the purpose of collection and whether any other entity may receive the data. If an AI vendor will process applications, that is worth saying.

A plain line in your applicant privacy notice could read:

We use [tool name] to help organise and summarise applications against the criteria in the job advert. A member of our HR team reads every application and makes every shortlisting decision. To ask about this, or to request a review, contact [role and email].

Your data protection lead should settle the exact wording.

Four safe ways AI can help with screening

Each prompt follows the four Cs we teach at FuKazee: Clarity, Context, Constraints and Customisation.

1. Write the criteria before you see any CV

From the job description below, write five to seven screening criteria that are essential to the job. For each, say what evidence in a CV would show it, and what would not count. Do not include age, sex, marital status, home area, school name or career gaps as criteria. JOB DESCRIPTION: [paste]

Check: the panel agrees the criteria first, and they match the advert.

2. Design an application form that collects less

Design a short online application form for a [job title]. Ask only for information needed to judge the essential criteria below. Explain in one line under each question why we ask it. Do not ask for date of birth, marital status, religion, ethnic group, ID number or a photo. CRITERIA: [paste]

Section 25 of the Act says personal data should be "limited to what is necessary".

3. Summarise an anonymised application against the criteria

Do this only in a tool your organisation has approved, after removing names and identifying details.

Below are our screening criteria and one anonymised application. For each criterion, quote the words in the application that relate to it, or write "not found". Do not give a score, a ranking or a recommendation. Do not guess anything the application does not say. CRITERIA: [paste] APPLICATION: [paste]

Check: open the original application and confirm every quote. The recruiter, not the tool, decides.

4. Test your process for bias with invented CVs

Make copies of one invented CV that differ in one detail only, such as surname, home county or a career break, and compare how your process treats them. Our HR training guide sets out this paired-CV test step by step, with ten questions to ask any vendor.

AI can also draft a respectful regret email or a structured interview guide, as long as a person has already made the decision those documents follow.

What AI must not do in screening

  • Reject, rank or shortlist on its own. A named person makes every decision.
  • Guess personal traits or punish gaps. Age, ethnic origin, pregnancy or health inferred from a name, school, county or career gap.
  • Reject CVs for "sounding like AI". Many honest applicants polish CVs with AI, as our guide for job seekers explains. Test skills in interviews and work samples.
  • Receive real CVs in a public tool. A CV is personal data and often holds sensitive data such as marital status. Section 49 says sensitive personal data may be processed outside Kenya only with consent and confirmation of appropriate safeguards. An online AI tool may process data outside Kenya, so check where yours does.
  • Keep CVs forever. Section 39 says personal data is kept only as long as reasonably necessary. Old chats full of applicants' details are a retention problem.

How FuKazee trains HR teams on this

FuKazee runs hands-on AI training for HR teams in Kenya. We never ask a team to bring a real staff file or a real CV into the room, and recruiters practise bias checks with invented paired CVs before anyone trusts a screening step. Training runs on-site anywhere in Kenya by arrangement, or live online. See what an HR programme covers, or start with the first 30 days of AI for an HR manager.

Questions people ask

Is it legal to use AI to screen CVs in Kenya?

As of November 2026 we know of no Kenyan law that bans it outright. Section 35 of the Data Protection Act restricts decisions based solely on automated processing that significantly affect people, and the Employment Act bars discrimination in recruitment. Keep a person making the decision and ask your lawyer about your own tool.

Can we auto-reject applicants who fail a yes or no question?

In our view, that is still a decision based solely on automated processing, even with no AI involved. Have a person check rejected applications, and get legal advice before relying on any auto-reject rule.

Do we need a data protection impact assessment?

Very possibly. Regulation 49 lists automated decisions with significant effects and the processing of sensitive personal data as high risk, and section 31 requires a DPIA before high-risk processing starts. Your data protection officer should decide.

What if a rejected applicant asks why?

Section 26 gives people the right to access their personal data and to know how it is used. If the decision was solely automated, section 35(3) also lets them ask you to reconsider. Record who reviewed each application.

Sources

All sources last checked 14 November 2026.

Where FuKazee fits in

FuKazee is a Nairobi training company that has trained more than 3,000 people since 2022. We help HR teams use AI in recruitment without losing the human judgement the law expects. Tell us how your team hires today and what you want AI to help with. Email lead@fukazee.com

Want guided support instead?

Plan AI training around your team's real work. Answer a few quick questions about your sector, team size and goals, and we will recommend the right programme and format. Sessions are built on your workflows and include a usage policy.

Rodney MainaFounder & Director, FuKazee

Rodney has run FuKazee's training since 2022 and still teaches inside every programme line: corporate sessions, the postgraduate cohort and the KCSE examiner webinars. First Class Honours in Journalism and Mass Communication, University of Nairobi.