Articles

EU AI Act: what UK HR teams need to do now (a practical checklist)

The EU AI Act is landing — and even though the UK isn’t copying it word-for-word, it will affect many UK employers in practice (through EU operations, suppliers, clients, and “best practice” expectations).

This guide is written for UK SMEs and HR teams who want a simple, low-admin way to reduce AI risk without killing innovation.

1. Start with an AI inventory (your single best first step)

You can’t govern what you can’t see. Create a simple list of:

• AI tools staff use (e.g., Copilot, ChatGPT, recruitment tools, HR platforms with AI features)
• what they’re used for (drafting, screening, analytics, employee monitoring, etc.)
• what data goes into them (especially any personal or confidential data)
• who “owns” each tool internally

2. Define “allowed” vs “not allowed” use cases (in plain English)

Examples most UK employers should allow (with human review):

• drafting policies and internal comms
• summarising long documents
• creating first-pass training content structures

Examples most UK employers should not allow without specialist oversight:

• AI producing conclusions in disciplinaries/grievances
• automated candidate scoring/shortlisting with no human oversight
• “productivity scoring” or monitoring that could create bias or employee relations risk

3. Put a verification step into your workflow (and make it normal)

A simple rule that works:

• AI can draft
• Humans approve

For anything that could end up in front of an employee, union rep, solicitor, regulator, or tribunal: verify facts, check tone, and keep the final signed-off version in your HR system.

4. Tighten your data rules (UK GDPR still applies)

Do not paste special category data or detailed employee case information into unapproved AI tools.

If you need AI support, anonymise inputs and keep prompts high-level.

5. Train managers and HR on “AI literacy” (lightweight but consistent)

Most AI risk comes from inconsistent use. Give managers a one-page “do/don’t” plus a short briefing covering:

• what to use AI for
• what not to use AI for
• what must always be checked
• what data must never be entered

6) Make recruitment AI “high attention”

Recruitment is one of the fastest areas where AI can accidentally introduce bias, unfairness, or poor candidate experience.

If you use AI in recruitment, ensure:

• clear criteria
• documented human oversight
• an audit trail of decisions (why someone was progressed/rejected)

PeopleFirstHR takeaway

You don’t need a big-budget AI compliance programme. You need visibility + basic rules + human oversight.