Unfair Dismissal When AI Influenced the Decision (UK, 2026): What Your Rights Actually Are

More UK dismissals now have an algorithm somewhere in the chain. A productivity score that ranks you against your team. Monitoring software that logs keystrokes, screen time, or calls and turns them into a "performance" number. An automated flag that lands you in a capability process. The technology has moved faster than most people's sense of what their rights are when it decides, or helps decide, that they lose their job.

This is a plain-English guide to where you stand in 2026 if AI shaped your dismissal. It is written for the person on the receiving end.

First question: did the AI decide, or just inform?

Everything turns on how much of a human was really in the decision.

If a manager looked at the AI's output, considered your side, had the authority to disagree, and made a genuine judgment, that is a human decision informed by a tool. Ordinary unfair dismissal law governs it.

If the AI's output was effectively the decision, and the "human review" was someone approving the number because overriding it was impractical, that is a solely automated decision. Since 5 February 2026, the UK GDPR (Articles 22A to 22D, brought in by the Data (Use and Access) Act 2025) gives you four safeguards over a significant solely-automated decision: information about the decision, the ability to make representations, human intervention from the employer, and the ability to contest the outcome. A dismissal is about as significant as a decision gets. The full regime is set out in our reference on automated decisions under Article 22A to 22D.

The rubber stamp is the thing to watch. A signature at the bottom of an AI recommendation is not meaningful human involvement.

Unfair dismissal: the fairness test still applies, and AI does not soften it

To dismiss you fairly, an employer needs a fair reason under section 98 of the Employment Rights Act 1996 (capability, conduct, redundancy, legal restriction, or some other substantial reason) and a fair procedure, judged against the ACAS Code. An AI score changes none of that. If anything it adds ways for the process to go wrong:

• You were never shown the score, or how it was produced. A process you cannot see is hard to call fair.
• You could not challenge the data. If the monitoring missed context (you were on approved leave, doing untracked work, or carrying a disability the tool did not account for), and there was no route to correct it, the procedure is weak.
• The reason does not really stand up without the algorithm. If the only evidence of poor performance is an opaque number, a tribunal can find the employer did not have reasonable grounds for its belief.

The qualifying period, accurately, in 2026. Ordinary unfair dismissal currently needs two years' continuous service. The Employment Rights Act 2025 cuts that to six months, but the change takes effect on 1 January 2027, so through 2026 the two-year rule still applies. From that date the cap on the unfair dismissal compensatory award (currently £123,543, or 52 weeks' pay if lower) is also removed.

The route that needs no qualifying period: discrimination

This is the part most people miss, and it is often the strongest card.

Discrimination protection is a day-one right. There is no qualifying period. So even if you have under two years' service and cannot bring an ordinary unfair dismissal claim today, you may still have a discrimination claim if the AI disadvantaged you because of a protected characteristic:

• A productivity tool that penalises slower work can disadvantage disabled workers, and can expose a failure to make reasonable adjustments.
• A model trained on past data can carry forward bias on age, race, or sex.
• Monitoring that infers health or other sensitive information brings its own problems.

Two things make this route powerful. Discrimination awards are uncapped, unlike the ordinary unfair dismissal award. And the burden can shift: once you show facts from which discrimination could be inferred, the employer has to prove the AI did not discriminate, which is hard when they cannot fully explain how the model reached its result.

There is also automatically unfair dismissal, which is likewise day-one. If you were dismissed for raising a genuine concern about the monitoring or the AI itself (a protected disclosure), or for asserting a statutory right, that can be automatically unfair regardless of service.

What you can demand to see

You are not limited to what the employer volunteers.

• Make a subject access request. You are entitled to the personal data held about you, which includes the monitoring and performance data behind the decision, and to meaningful information about the logic of any automated decision-making and its likely consequences for you.
• Use the automated-decision safeguards. Where the decision was solely automated, require human intervention and contest the outcome in writing.
• Ask for the score and its basis at the appeal. Put the request in the appeal letter. If the employer cannot produce a clear, challengeable basis, that strengthens your position.
• Keep your own record. Dates, the process followed, what you were and were not shown, and anything that explains the data (leave, untracked work, a disability or adjustment).

For the wider picture of what employers are and are not allowed to do with monitoring in the first place, see our guide to workplace surveillance and AI monitoring.

When to take advice

Some situations are worth a specialist's eye early, because the time limits in employment law are short (an employment tribunal claim is usually three months less one day from the dismissal, after early conciliation through ACAS).

Take advice promptly when:

• An AI score or monitoring output was central to your dismissal or capability process.
• You have under two years' service but believe bias or a protected characteristic was involved (the discrimination route may still be open).
• You have a disability the monitoring did not account for, or asked for an adjustment that was refused.
• You were dismissed after raising concerns about the monitoring or the AI.
• The employer cannot or will not show you the basis for the decision.

---

Michael K. Onyekwere is a common law qualified lawyer and a CIPP/E-certified data protection practitioner, writing on the overlap between employment law and AI. This is educational and current as at 22 June 2026. It is not legal advice and not a substitute for advice from a qualified solicitor on your specific facts; for your matter, instruct a regulated practitioner, and mind the short tribunal time limits.

Part of the Janus Employment Law cluster and the AI and your rights theme. See also: Automated decisions under Article 22A to 22D, Workplace surveillance and AI monitoring, AI rejected your job application.