AI in the workplace: why HR must take control

A recruiter who summarises CVs with a free tool. A team leader who lets AI phrase a performance review. Nobody has banned it, nobody has arranged it. AI in the workplace stopped being a thing of the future a long time ago. The question is not whether your employees use AI, but whether someone stays in control. And that is an HR question.

Almost half of working people are not aware of any AI policy, or do not know what it says. Without agreements, everyone picks their own tool. Then you do not know where personal data ends up, whether a text is correct and whether everyone gets the same opportunities.

AI makes work faster, not always better

The first experiments are clear. In a study with 758 BCG consultants, people using AI completed 12% more tasks, 25% faster and with 40% higher quality, but only on tasks that AI handles well. On tasks outside that frontier, they were 19 percentage points less likely to give the right answer: they accepted the AI output too easily (Dell’Acqua et al., 2023).

So experience becomes more valuable, not less. Someone has to judge the outcome, and starters still have to learn the trade. Read more about how AI changes the employee experience in this blog.

Which rules already apply?

The European AI Act is coming into force in stages:

• Since 2 February 2025 AI that recognises employees’ emotions in the workplace is prohibited, except for medical or safety reasons. And you make sure employees know what AI can and cannot do.
• Since 2 August 2026 people must know that they are dealing with AI, for example with an HR chatbot.
• From 2 December 2027 strict requirements apply to AI that helps decide on recruitment, selection, appraisal, promotion or dismissal.

And in the Netherlands it has long been the case that a decision with major consequences may not be taken by a computer alone (GDPR, Article 22), and the works council must approve systems that can monitor behaviour or performance (Dutch Works Councils Act, Article 27). This is not legal advice; check the details with your lawyer or data protection officer.

When does AI help decide about people?

Not every AI in HR falls under the strict rules. The question is always: does it influence a decision about someone’s job or career? Drafting a job ad or summarising a policy is usually not high risk. Filtering or ranking CVs, assessing candidates and monitoring performance is. The vendor supplies the system, but you are responsible for how you use it: human oversight, keeping logs and informing people in advance.

Employees mainly want clarity

Employees are curious and often positive. But 84% worry about privacy with AI, only 37% trust AI to give correct answers and 16% feel pressure to use AI. Ask them what they need. Anonymously, per team: what am I allowed to do, what do I need, what changes for me? Do not measure it with AI: emotion recognition in the workplace is prohibited.

Seven things to arrange now

1. List which AI is already in use, including in your HR systems.
2. Teach people what AI can and cannot do, suited to their role (mandatory).
3. Write down the house rules: which tools are allowed, and which data must never go into a public tool?
4. Assess your HR systems before December 2027 (mandatory).
5. Involve the works council and your people early (mandatory).
6. Look at work and learning: which tasks change, and how do starters learn the trade?
7. Ask how it lands: clarity, pressure and opportunities per team.

Get started

With the Learning and development module in Deepler Research you measure development needs per team and role, anonymously and at group level. The whitepaper includes a quick scan, an inventory and ten questions to ask before you sign with a vendor.

Sources

• CBS (2025). Use of AI by companies, December 2025.
• PwC (2025). Hopes and Fears 2025, Dutch results, 1,257 employees.
• KPMG and Ipsos I&O (2025). AI Vertrouwensmonitor 2025, 1,013 Dutch people.
• Dell’Acqua, F. et al. (2023). Navigating the jagged technological frontier. Harvard Business School Working Paper 24-013.
• Brynjolfsson, E., Li, D. and Raymond, L. (2025). Generative AI at work. Quarterly Journal of Economics, 140(2).
• Regulation (EU) 2024/1689 (AI Act) and Regulation (EU) 2026/1744 (Digital Omnibus on AI).