Use AI for · HR · 26 Sep 2026
Use AI for
AI in HR

AI in HR:what it can take on,what stays with people

AI drafts job adverts, prepares interviews and builds onboarding plans. This guide shows how to approach each task, which instructions work, and where the EU AI Act, equal treatment law, data protection and works councils set limits.

Get in touch
On this page
Enquiry
Nikolai Schöbel und Jeremias Burger, Co-Founder Scalableloops

Let's talk about your project.

First we check whether the project fits your business model. Then you get a proposal with phases and effort.

Discuss AI in HR or call: +49 151 1576 5566
Use AI for · HR

Using AI in HR mostly means handing over writing and preparation while keeping decisions about people in human hands. A language model such as ChatGPT, Claude, Gemini or Microsoft Copilot can write the first draft of a job advert, suggest interview questions and turn your documents into an onboarding plan. As soon as a system filters applications or assesses people, strict rules apply under the EU AI Act, equal treatment law, data protection law and, in Germany, works council law.

In brief
  • AI works well for job adverts, interview guides, onboarding plans and answers to recurring questions from new starters.
  • AI systems that screen or filter applications or evaluate candidates are listed in Annex III of the EU AI Act and are in principle classed as high-risk.
  • Under the German General Equal Treatment Act, job adverts must be free of discrimination, even when an AI wrote the draft.
  • In Germany, the works council must be informed about AI use and, depending on the system, has a say in it.
Published 26 Sep 2026Nikolai Schöbel and Jeremias Burger10 min read
Nikolai SchöbelJeremias Burger

Nikolai Schöbel and Jeremias Burger

Co-founders of Scalableloops GmbH. Nikolai Schöbel leads online marketing and AI strategy, Jeremias Burger the AI architecture. Both build AI systems and train teams on them in their own agency work.

On this page
  1. Where does AI make sense in HR?
  2. How do you write a job advert with AI?
  3. Can AI pre-sort applications or rate candidates?
  4. How can AI help new starters settle in faster?
  5. What does the GDPR mean for applicant and employee data?
  6. When does a German works council have a say in AI?
  7. Which mistakes should HR teams avoid with AI?
  8. Frequently asked questions
  9. How to get started with AI in HR
  10. Where the information on this page comes from
Overview

Where does AI make sense in HR?

A large part of HR work is text: adverts, invitations, rejections, guides, onboarding plans and answers to the same questions again and again. That is exactly where an AI assistant takes work off your desk. You get a draft, then check and adapt it.

It gets sensitive where AI stops writing and starts judging people: pre-sorting applications, rating candidates, assessing performance. Separate rules apply to these tasks. The table shows where to look more closely.

The tool matters less. What counts when you bring an assistant into your business is covered in our article on the AI assistant for business.

TaskWhat AI can doWhat stays with peopleLegal points
Job advertFirst draft, versions for different channels, translationDefining the requirements, checking and approving the textNon-discriminatory advert under the German General Equal Treatment Act (AGG)
Interview guideQuestions based on the job requirements, a scoring templateChoosing questions, running the interview, forming a viewNo questions linked to characteristics protected by the AGG
Screening applicationsOnly with a system designed and assessed for this purposeThe selection decisionAnnex III EU AI Act, Article 22 GDPR, works council
OnboardingOnboarding plan, welcome email, answers from your own documentsKeeping content current, naming contact peopleData protection for employee data
Job adverts

How do you write a job advert with AI?

A good job advert does not come from a single sentence. Give the AI what only you know: tasks, requirements, working hours, location, what makes your business different and how you usually write. Our article on using AI for writing explains how to build instructions in general.

An instruction could look like this: “Write a job advert for a part-time accounts assistant, 25 hours a week, based in our Munich office. Tasks: accounts payable, payment reminders, preparing documents for our tax adviser. Requirements: commercial training, confident with accounting software. Use a friendly but professional tone. Structure: short intro, tasks, requirements, what we offer, contact. Avoid clichés such as ‘fast-paced environment’.”

Then comes the most important step: checking for discrimination. In Germany, section 11 of the General Equal Treatment Act (AGG) prohibits advertising a job in breach of the ban on discrimination. The protected characteristics in section 1 are race or ethnic origin, sex, religion or belief, disability, age and sexual identity. Under section 6, applicants already count as employees for the purposes of the Act.

A language model reuses wording from the texts it was trained on, including phrases that target an age group or a gender, such as “young team” or a job title that only names one gender. The Federal Anti-Discrimination Agency recommends adverts that are neutral as to these characteristics and gender-neutral job titles. Check every draft, or ask the AI: “Check this advert for wording that could target age, sex or origin under German equal treatment law and suggest neutral alternatives.”

For other languages, DeepL or the language models above produce a draft. Have it checked by someone fluent in the language; more on this in our article on using AI for translation.

Selection

Can AI pre-sort applications or rate candidates?

This is where the EU AI Act draws a clear line. Point 4 of Annex III lists AI systems intended to be used for the recruitment or selection of people, in particular to place targeted job adverts, to analyse and filter applications and to evaluate candidates. Such systems are in principle high-risk AI. The same applies to systems that monitor and evaluate employees’ performance and behaviour.

Article 6(3) allows exceptions, for example where a system only performs a narrow procedural task or a preparatory task and does not materially influence the outcome of a decision. The exception never applies if the system profiles natural persons. Have your specific use checked legally.

The obligations for high-risk systems include the duty of employers under Article 26(7) to inform workers’ representatives and affected workers before a high-risk system is used at the workplace. When these obligations start to apply was reset by the amending Regulation (EU) 2026/1744, known as the Digital Omnibus, so check the current position before introducing such a system. Already banned since 2 February 2025 are AI systems that infer the emotions of people in the workplace, apart from narrow medical or safety reasons.

Separately, Article 22 GDPR gives everyone the right not to be subject to a decision based solely on automated processing that produces legal effects for them or similarly significantly affects them. So do not let a system decide a rejection on its own.

A well-known case shows how quickly a system picks up bias. Amazon trained a CV rating tool on ten years of applications. Because most of them came from men, it downgraded CVs containing the word “women’s”. Amazon dropped the project. A general AI chat is not a selection tool. Use it to sharpen requirements and prepare interviews, not to rank people.

Onboarding

How can AI help new starters settle in faster?

Onboarding offers high benefit at low risk, as long as nobody is being assessed. The AI works with your own material: job description, handbook, checklists.

An onboarding plan might start like this: “Here is the job description for our new project assistant and our onboarding checklist. Create a plan for the first four weeks, organised by week, with contact people from the list, dates for feedback conversations and the software she needs to learn. Mark everything that has to be done before day one.”

An assistant that answers new starters’ questions about holiday requests, travel expenses or working hours from your internal policies is just as useful. It should only answer from approved, current documents and refer to a named contact when unsure. You can also turn onboarding conversations into minutes, as described in our article on using AI for meeting minutes.

If you run such an assistant for applicants on your careers page, Article 50 of the EU AI Act has applied since 2 August 2026: people must be told they are interacting with an AI. Details are in our article on chatbot AI disclosure under Article 50.

Data protection

What does the GDPR mean for applicant and employee data?

CVs, references and personnel files contain personal data. They do not belong in a personal AI account. With personal ChatGPT accounts such as Free and Plus, training on your inputs is switched on until you turn it off in the data controls. For the business offerings ChatGPT Business and Enterprise, OpenAI says it does not use your data for training by default, and only business customers get a data processing agreement.

In its guidance on AI and data protection, the German Data Protection Conference (DSK) recommends that employers provide accounts and devices for work use instead of letting staff use personal accounts. For cloud services you will usually need a contract under Article 28 GDPR.

Most tasks need no personal data. A job advert, an interview guide or an onboarding plan can be written without names. Where you work with real documents, for instance when summarising interview notes, leave out names and contact details or use a business account with a contract. If you want to keep particularly sensitive data in house, our article on local AI for business shows a way to do it.

Works council

When does a German works council have a say in AI?

If your business in Germany has a works council, involve it early. The Works Constitution Act (BetrVG) mentions AI explicitly in several places. Under section 90(1) no. 3, the employer must inform the works council about planned work procedures and workflows, including the use of artificial intelligence. Section 95(2a) makes clear that co-determination on selection guidelines also applies when AI is used to draw them up.

The strongest lever is section 87(1) no. 6: it gives the works council a co-determination right over technical systems designed to monitor employees’ behaviour or performance. How far that goes with AI was shown by the Hamburg Labour Court on 16 January 2024 (24 BVGa 1/24). It found no co-determination right where staff used ChatGPT in the browser with personal accounts, because no software was installed on company hardware and the employer had no access to usage data.

That is no all-clear. With a company-provided tool whose usage data the employer can access, the position may differ. A works agreement on AI use gives both sides clarity.

Practice

Which mistakes should HR teams avoid with AI?

Most problems come from missing rules rather than from the technology. Settle these points before you start:

  1. 01

    01

    Taking drafts at face value

    AI text sounds finished but can contain wrong details or wording that is risky under equal treatment law. A person checks every advert and every message to applicants.

  2. 02

    02

    Personal data in personal accounts

    CVs and personnel files only go into tools with a business account and a data processing agreement, and it is better still to leave names out.

  3. 03

    03

    Using a general chat to select people

    A chat that ranks CVs by suitability takes on a task for which the EU AI Act sets strict requirements. Selection stays with people.

  4. 04

    04

    Involving the works council too late

    Informing it only after roll-out risks conflict and delay. The duty to inform under section 90 BetrVG already covers the planning stage.

  5. 05

    05

    Not preparing the team

    Since 27 July 2026, Article 4 of the EU AI Act has applied in amended form: businesses using AI must take measures that support the development of AI literacy, without having to guarantee a specific level for each person. For HR this means knowing which data may go in and where the limits are. More in our article on the AI literacy obligation under Article 4.

Frequently asked questions

Frequently asked questions

Is AI in recruitment banned under the EU AI Act?

No. But AI systems that screen or filter applications or evaluate candidates are in principle high-risk under Annex III of the EU AI Act. Special obligations apply to them, and the Digital Omnibus has reset when they start. Using AI to recognise emotions in the workplace has been banned since 2 February 2025, with narrow exceptions.

Can I write a job advert with ChatGPT?

Yes. A job advert does not need any personal data. Before publishing, check that the draft does not favour characteristics such as age or sex, which German equal treatment law prohibits in job adverts. Responsibility for the text stays with you.

Can I upload CVs to ChatGPT?

Not with a personal account. CVs contain personal data, so you need a business account with a data processing agreement and training switched off. Do not use AI to rank or reject applicants either, because Article 22 GDPR protects people against decisions based solely on automated processing.

Does the works council have to agree before HR uses AI?

It depends on the system. Under section 90(1) no. 3 BetrVG you must inform it at the planning stage. It has co-determination rights under section 87(1) no. 6 if the system can monitor behaviour or performance, and under section 95(2a) for selection guidelines drawn up with AI.

Does the HR team need AI training?

Since 27 July 2026, Article 4 of the EU AI Act has required businesses that use AI to take measures that support the development of AI literacy. No certificate is required. For HR, the focus is on data protection, equal treatment and the limits of AI in decisions about people.

What now

How to get started with AI in HR

  1. 01

    Sort the tasks

    Separate writing tasks such as adverts, guides and onboarding plans from tasks where decisions about people are made. Start with the writing tasks.

  2. 02

    Settle tool and contract

    Use a business account with a data processing agreement and training switched off. Keep names and personal data out of your instructions wherever possible.

  3. 03

    Involve the works council

    Inform the works council at the planning stage and put rules for AI use in writing, ideally in a works agreement.

  4. 04

    Prepare the team

    Train your HR team on what is allowed, where AI gets things wrong and who signs off. Scalableloops supports you with AI training based on your own tasks.

AI takes a lot of writing off the HR desk, from the job advert to the onboarding plan. Decisions about people stay with people, and that needs clear rules, the right contract and a works council involved from the start.

or call: +49 151 1576 5566

Further reading
Sources

Where the information on this page comes from

Projekt-Detail

Problem

Our solution

Our role

Highlights

    Got a project in mind?

    We reply personally. First a use-case check, then an architecture proposal.

    Start your inquiry