Try it free
  1. Home
  2. Knowledge
  3. AI recruiting in Denmark: what AI can do and the rules

AI recruiting in Denmark: what AI can do and the rules

Updated

AI recruiting means using artificial intelligence in the steps of recruiting: reading and writing job ads, finding candidates, sorting applications and drafting messages. In 2025, 42% of Danish companies with at least 10 employees used AI, up from 15% in 2023, according to the Danish Agency for Digital Government and Statistics Denmark. Here's where AI helps, what a person still has to decide, what the rules require, and what to ask a vendor. This article is practical guidance, not legal advice.

Where AI helps

AI is best at what takes time without needing your judgment: reading a lot of text, finding patterns and writing drafts.

  • The job ad and the requirements. AI can read a job ad and pull out the requirements: function, seniority, location, years of experience and skills, and which are must-haves and which are nice-to-haves. A vague ad makes a vague search, so this is where quality starts. Read How to write a job ad a search can use.
  • Sourcing. AI can turn the requirements into a search with title variants in Danish and English, search a database of profiles and rank the ones that fit. That's the part that used to take hours of Boolean strings. More in AI sourcing in Denmark and the Nordics.
  • Applications. An ATS with AI can sort incoming applications against the requirements, so you read the most relevant first.
  • Messages. AI can draft a first message from the candidate's career and the role, and sort the replies. You read and edit before anything is sent.
  • Scheduling and overview. AI can find interview times and sum up where each role stands.

What a person must decide

AI can sort, but a person has to decide who moves on, and be able to explain why.

  • The decision. Article 22 of the GDPR gives people the right not to be subject to a decision based solely on automated processing that significantly affects them. Recital 71 names e-recruiting without any human intervention as an example.
  • Actually looking. The AI Act itself names the risk of relying on a system's output too much. Approving a ranked list without looking at it isn't oversight. Look at the evidence for each requirement, and bring back a candidate the system screened out when you think it got it wrong.
  • What the data doesn't show. Languages, education and much else aren't always in the data a tool searches. Missing data should never exclude anyone, and what needs checking should be an open question, not a no.

The rules in short

The EU AI Act. An AI system for recruiting or selecting people, for example to analyse and filter applications and evaluate candidates, is high-risk under Annex III, point 4(a). Since 2 February 2025, AI that infers emotions in the workplace has been banned, and after the Digital Omnibus, the high-risk rules apply to recruiting from 2 December 2027. If you use a vendor's tool, you're the deployer, with duties to use it as instructed, keep human oversight, keep the logs and tell candidates about it. Read it all in The EU AI Act for recruitment agencies.

GDPR. You need a legal basis, and when you source candidates who haven't applied, you owe them the information in Article 14, at the latest in your first message. Read GDPR and sourcing passive candidates.

What to ask the vendor

  1. Data. What data does the tool search, and how many relevant profiles are there in Denmark?
  2. Evidence. Can you see what each assessment rests on?
  3. The decision. Does a person make every decision, and is it recorded?
  4. What the model sees. Does the AI model get names, photos or details such as gender and age?
  5. Where data is processed. Is there a data processing agreement and a public list of sub-processors with their locations and transfer mechanisms?
  6. Deletion. When are candidates you don't use deleted?
  7. The AI Act. How will the vendor meet the high-risk requirements before 2 December 2027?
  8. Price. Do you pay per user, per search or per hire, and in which currency?

How Navigent uses AI

Navigent reads the job ad and sets filters you can remove. One search fetches about 1,000 unique candidates, the best 200 or so make the longlist, and up to about 50 recommended are assessed requirement by requirement, with evidence from the jobs each assessment rests on. What search data can't answer is marked "To clarify". You make every yes and no decision, and every decision is saved. The AI model never sees names, photos, email addresses or phone numbers.

Try it with your own job ad in the free Requirements profile tool, with no sign-up, or see How it works.

Frequently asked questions

What is AI recruiting?

AI recruiting is the use of artificial intelligence in the steps of recruiting: reading job ads, finding and ranking candidates, sorting applications and drafting messages. A person still makes the decisions.

Yes, when the rules are followed. The GDPR already applies, AI that infers emotions in the workplace has been banned since 2 February 2025, and the EU AI Act's high-risk requirements apply to recruiting from 2 December 2027.

Will AI replace the recruiter?

No. AI can find, sort and draft, but a person has to assess the candidates and make the decision. Article 22 of the GDPR gives people the right not to be subject to a decision based solely on automated processing.

What does AI recruiting software cost?

It depends on the tool. Juicebox costs USD 119 per user per month billed monthly, SeekOut Sourcing Core USD 179, and hireEZ starts at USD 494 a month for a recruiter working alone. Navigent costs DKK 1,499 per user per month excluding VAT on Basis, and you can try it free: 2 free searches, 1 more per colleague, up to 5, no card. See Candidate sourcing tools.

Sources