FCA authorised · FRN 724952 0117 325 0027Quote & buy →
AI insurance
AI bias claims arise when an algorithm used in hiring, lending, pricing or access to services treats people less favourably because of a protected characteristic, or disadvantages them indirectly. They matter because UK employment discrimination awards have no cap, and the business using the tool stays answerable for it even when a vendor built it.
Part of: AI insurance UK
In short
Under the Equality Act 2010, an employer or service provider is liable for discrimination caused by an AI tool it uses, and indirect discrimination can arise from a neutral-looking screening rule that disadvantages a protected group. Employment tribunal compensation for discrimination has no upper limit. Vendors face claims mainly through their contracts, though Mobley v Workday in the US shows direct claims are possible. Since 5 February 2026 the Data (Use and Access) Act 2025 has reshaped UK rules on automated decisions. For AI vendors and consultants, professional indemnity is usually the main cover; an employer's own discrimination claims fall under employment practices liability insurance.
The Equality Act 2010 does not mention AI, but its tests apply to decisions made with it:
Section 39 stops an employer discriminating "in the arrangements A makes for deciding to whom to offer employment", so job applicants screened out by software can claim. Section 29 applies the same duty to businesses providing services to the public, which reaches AI used for lending, tenancy checks, pricing or customer verification.
The burden of proof helps claimants. Under section 136, if there are facts from which a tribunal could decide discrimination occurred, it must find so unless the respondent shows it did not. An employer that cannot explain how its tool ranks candidates may struggle to discharge that burden. See explainable AI and insurance.
GOV.UK says there can be limits on what an employment tribunal awards, but "there's no limit in cases of discrimination" (GOV.UK guidance). Awards can include financial loss and injury to feelings, and defence costs are often significant even when a claim fails.
AI adds a scale problem. A human recruiter who discriminates affects the candidates they see. A screening model applies the same rule to every applicant, so one flawed criterion can produce many claims from one cause. That affects how a policy's aggregation and limit of indemnity work: see aggregation of claims in PI.
In the UK, the employer or service provider using the tool is the natural defendant: the decision, and the arrangements behind it, are its own. Where a supplier acts as its agent, section 109 treats anything done by the agent with the principal's authority as also done by the principal, whether or not the principal knew or approved.
The vendor's exposure usually comes through contract. The employer settles or loses the claim, then seeks to recover from the vendor for breach of warranty, negligence or under an indemnity. Section 112 also makes it unlawful to knowingly help another person contravene the Act, although "knowingly" sets a high bar.
| Party | How a bias claim reaches them | Cover that may respond, subject to the policy terms |
|---|---|---|
| Employer using an AI hiring tool | Tribunal claims by applicants and employees | Employment practices liability (a separate policy) |
| Business using AI to decide on customers | County court discrimination claims; data protection complaints | Depends on the wording; check general liability and cyber policies for discrimination and AI exclusions |
| AI vendor or developer | Contract and negligence claims from the customer; possible direct claims | Professional indemnity (technology errors and omissions) |
| AI or HR consultant who chose or configured the tool | Negligent advice or configuration claims by the client | Professional indemnity |
The most watched case is in the US. In Mobley v Workday, Inc. (N.D. Cal., No. 23-cv-00770), a federal court on 12 July 2024 refused to dismiss disparate impact claims on the theory that the vendor could be liable as the employers' agent, because employers had allegedly "delegated their traditional function of rejecting candidates or advancing them to the interview stage" to its software. The court rejected the separate theory that the vendor was an employment agency. On 16 May 2025 it preliminarily certified a nationwide age discrimination collective, and on 22 June 2026 it allowed further claims to continue while dismissing others. These are pre-trial rulings, not findings of liability, and they do not bind UK courts. They show a vendor can be pulled into claims directly.
In the UK, an Uber Eats courier who alleged that the app's facial recognition checks were racially discriminatory, and who was removed from the platform after a failed check, received a financial settlement announced in March 2024. The case was supported by the Equality and Human Rights Commission (EHRC announcement). The announcement did not give the terms or say whether liability was admitted.
A data protection failure and a discrimination claim often arise from the same facts. See GDPR and insurance for technology companies.
For a business that builds, sells or configures AI used in decisions about people, professional indemnity is usually the main cover. It may respond when a client claims the tool was negligently designed, tested or set up and seeks its own losses, including what it paid out to candidates, subject to the policy terms. Some PI wordings exclude discrimination or contractual liability, so check how yours treats a client's claim to recover those sums. See does PI cover AI mistakes.
For an employer using an AI hiring tool, claims by its own applicants and staff usually fall under employment practices liability insurance. That is a separate cover and is not one of the four options in Apex's AI insurance scheme; see also PI vs EPL for HR consultants.
Cyber cover may respond to the data protection side, such as an ICO investigation, subject to the policy terms.
Keep evidence: see what to document about your AI use.
Apex has an AI insurance scheme with an insurer that names your specific AI work, such as a screening, scoring or decision tool you build or configure, in the policy as insured activity. You choose from professional indemnity, cyber, product and public liability, and IP infringement cover, and cover for US exposure is available, subject to the insurer's underwriting and the policy terms. To quote, we need a precise description of your AI work, your address, your turnover (including how much comes from the USA and Canada) and the limit of indemnity you want; you can start an AI insurance quote online in a few minutes. See AI insurance UK, US cover for UK AI firms and AI contract clauses and indemnities.
Yes. Under section 39 of the Equality Act 2010 an employer must not discriminate in the arrangements it makes for deciding whom to employ, including arrangements that rely on software. Section 109 treats acts done by an agent with the employer's authority as the employer's own. Rejected applicants can bring employment tribunal claims.
Sometimes. In the UK the employer is usually the defendant and then looks to recover from the vendor under the contract or in negligence. In the US, Mobley v Workday allowed claims that a screening vendor acted as the employers' agent to proceed in 2024. Vendor contracts often cap liability, so read them closely.
No. GOV.UK says there can be limits on employment tribunal awards, but there is no limit in discrimination cases. Awards can include lost earnings and injury to feelings. Because an AI tool applies the same rule to every candidate, one flawed criterion can lead to many claims.
It can, for the business that built, sold or configured the tool, if a client claims its losses from a negligent design or set-up, subject to the policy terms and any discrimination or AI exclusion. An employer's claims from its own applicants usually need employment practices liability insurance, which is a separate policy.
From 5 February 2026 it replaced Article 22 of the UK GDPR with Articles 22A to 22D. Solely automated significant decisions are generally allowed with safeguards: informing people, letting them make representations, obtain human intervention and contest the decision. Stricter limits apply when special category data is involved.
Tell us what your AI work is, where you are based, your turnover and the limit you need. It takes a few minutes online, and a broker will come back to you with options. Or call 0117 325 0027.
Start an AI insurance quote Request a callbackApex Insurance Brokers Limited is authorised and regulated by the Financial Conduct Authority. Registered in England and Wales, company number 07014570. This page is general information, not advice on your individual circumstances, and it does not guarantee that cover will be available or on what terms.
