What AI can and can't do in UK HR
Reviewed by Mellow Editorial Team, HR & payroll content team
AI tools can genuinely reduce administrative burden in UK HR — drafting job descriptions, summarising documents, flagging anomalies in payroll data. They cannot make legally compliant employment decisions, replace human judgement where statutory rights are at stake, or take responsibility when something goes wrong.
Where AI adds real value in HR admin
The most reliable use cases are the ones where AI is doing drafting or organising work that a human then reviews.
Writing job descriptions, interview question banks, onboarding checklists and policy summaries are all reasonable tasks to hand to an AI tool. The output needs editing, but the time saving is genuine. Similarly, AI can scan large documents — an employment contract, a staff handbook — and pull out relevant clauses, which is useful when a manager needs a quick answer at short notice.
In payroll, AI-assisted tools can flag unusual patterns: a spike in overtime, a missing pay element, an employee whose hours changed but whose rate did not. That kind of anomaly detection is valuable precisely because payroll errors compound quickly — a missed RTI Full Payment Submission to HMRC, for example, can trigger penalties and employee complaints simultaneously.
AI can also support recruitment screening by sorting CVs against criteria you define. This is useful for volume hiring. But it requires careful setup and ongoing oversight — more on that below.
What AI cannot do
AI cannot hold legal responsibility. If an AI tool produces an incorrect redundancy calculation and you act on it, the liability is yours as the employer. Employment tribunals will not accept "the software told me to" as a defence.
AI also cannot interpret context the way an experienced HR professional can. It does not know that the employee flagged as frequently absent is currently undergoing cancer treatment, or that the team's low engagement scores follow a specific management change. It works with the data it is given, and UK employment situations often turn on facts and nuance that are not in any system.
Statutory rights are another hard limit. Auto-enrolment pension contributions — employer minimum 3%, employee 5% of qualifying earnings — must be calculated and applied correctly regardless of what any tool suggests. The personal allowance, tax bands, employer National Insurance at 13.8%: these are fixed by law. An AI that gets these wrong, even by a small margin, creates real compliance exposure. Always verify statutory figures against HMRC guidance rather than trusting AI-generated numbers.
The Employment Rights Act 2025 strengthened day-one rights for employees, including stronger protections around unfair dismissal and flexible working requests. AI tools trained on older data may not reflect these changes. Any AI-generated policy or letter touching on statutory entitlements needs a human check against current law.
The bias and discrimination risk
This is the area where the gap between AI capability and legal requirement is sharpest.
Algorithmic hiring tools have been shown repeatedly — in research and in tribunal cases internationally — to perpetuate historical bias. If your past hires skewed toward a particular demographic, an AI trained on that hiring data will tend to favour similar profiles. Under the Equality Act 2010, indirect discrimination is unlawful even when unintentional. "We used an algorithm" does not create a defence.
The ICO and the Equality and Human Rights Commission have both published guidance on automated decision-making in employment contexts. The key principle: no significant decision about an employee or candidate — shortlisting, promotion, dismissal — should be taken by an automated process without meaningful human review. This is also a requirement under UK GDPR, which gives individuals the right to object to solely automated decisions that have a legal or similarly significant effect on them.
Practical rules for using AI tools responsibly
A few principles that hold across most AI use cases in HR:
Keep a human in the loop for anything consequential. Drafting is fine to automate. Decisions are not.
Check outputs against current legislation. AI tools have training cutoffs and may not reflect recent statutory changes. If a letter or calculation touches on legal entitlements — sick pay, family leave, notice periods — verify it independently.
Document your process. If you use AI to assist in a recruitment or disciplinary process, record what it produced, what you reviewed, and what decision you made. This protects you if a decision is later challenged.
Tell employees when AI is involved. Transparency is both a legal obligation under UK GDPR and good practice. If AI has assisted in evaluating an application or assessing performance data, employees have a right to know.
Audit periodically. AI tools drift. A screening tool that seemed fair when you set it up may be producing skewed outputs six months later. Regular spot-checks on outcomes — not just processes — help catch this.
The honest bottom line
AI in HR is genuinely useful as a productivity tool for low-stakes, reversible tasks. It becomes a liability when it is treated as a decision-maker rather than a drafting assistant. The employers who use it well are the ones who stay clear on that distinction and build human review into every process that touches statutory rights or individual employment outcomes.
---
Run HR and payroll in United Kingdom with Mellow
Mellow brings HR, payroll and 12 AI agents into one platform — built to handle United Kingdom properly, with payroll included, from £4 per employee per month. The AI agents don't just answer questions; they generate contracts, run cost estimates and draft letters for you.
- United Kingdom payroll software
[Start a free trial →](/register)