By the Editors · Published 3 July 2026 · Last updated 3 July 2026 · 12 minute read

Is my AI agent covered by employers liability insurance?

Short answer: no. Employers liability insurance is a statutory policy built around human employees, and an AI agent does not meet the legal definition of one, no matter how much of your operation it runs. This guide explains what employers liability actually covers, why the "insured like an employee" language around ElevenLabs's February 2026 AIUC-1 policy does not mean what it sounds like, and which policies SME operators should be checking instead.

Section 1: Why the question comes up

The question "is my AI agent covered by employers liability insurance" usually starts from a reasonable place. Founders and operations leads increasingly describe their AI agents in the same language they use for staff: the agent "works" the support queue, "handles" onboarding calls, "covers" a shift. If an agent is doing a job a person used to do, it seems logical to ask whether the insurance that covered the person now covers the agent.

The coverage does not follow the job description. Insurance policies are written against legal categories, not job functions. Employers liability insurance is written specifically against the legal category of employee, and an AI agent, no matter how central to the business, does not fall into that category under any insurance or employment framework currently in force in the EU, the UK, or the US.

Section 2: What employers liability insurance actually covers

Employers liability insurance is a statutory line in most European jurisdictions, required by law wherever a business employs staff. In the UK, the Employers' Liability (Compulsory Insurance) Act 1969 requires almost all employers to carry at least GBP 5 million of cover. The policy responds to claims brought by an employee against their employer for bodily injury, disease, or psychiatric harm suffered in the course of employment, where the employer is found to have been negligent in its duty of care.

Three elements are load-bearing in that definition, and an AI agent fails all three. First, the claimant must be an employee, a legal status created by an employment contract or equivalent statutory relationship. An AI agent has no employment contract and no legal personhood that would allow it to hold one. Second, the harm must be bodily injury, disease, or recognised psychiatric injury. An AI agent cannot suffer any of these. Third, the claim runs from the injured party to the employer, not the other way round. Employers liability protects the business against claims brought by its own staff, not against claims the business's tools or systems cause to third parties.

None of this means an AI agent creates no liability. It means the liability an AI agent creates sits in a different part of the insurance programme entirely, specifically in the policies that respond to harm the business causes to customers, clients, or other third parties.

Section 3: What ElevenLabs actually did in February 2026

In February 2026, ElevenLabs became the first company reported to hold an insurance policy for its AI voice agents underwritten against the AIUC-1 certification standard published by the AI Underwriting Company. Coverage was widely described in press coverage as insuring the agents "like any other employee," and that framing has understandably led some SME operators to ask whether an equivalent employers liability style product now exists for AI agents generally.

It does not, at least not in the form the phrase implies. The AIUC-1 backed policy ElevenLabs holds is a standalone AI liability product. It prices coverage against an independent audit of the agent's behaviour, in the same way AIUC's broader model combines certification and audit with insurance for hallucination-driven loss, data leakage, harmful outputs, and faulty tool actions. The "like an employee" language is a description of the company's risk posture, treating the agent's actions with the same seriousness as it would treat a human employee's actions, not a statement that the agent has been added to a payroll or an employers liability schedule. No carrier, including AIUC, has proposed extending statutory employers liability cover to non-human agents, because the legal category the policy is built on does not admit them.

What the ElevenLabs case does demonstrate, and what is genuinely useful for SME operators to take from it, is that a named, audited, purpose-built AI liability policy is now a real, bought, in-force product category, not a theoretical one. That is the direction operators should be looking for coverage, not toward stretching an existing employers liability schedule to cover something it was never built for.

Section 4: The policies that actually respond

For an SME running an AI agent, the realistic coverage map looks like this. Professional liability, also called Errors and Omissions, is the closest existing policy type for agents that advise, recommend, book, or communicate with clients. It was written before autonomous agents existed, so the wording on AI-generated outputs is often ambiguous, but it is the policy most likely to have some bearing on a claim arising from an agent's professional-grade mistake.

Technology Errors and Omissions extends that logic specifically to technology-enabled service delivery, and is the better fit for businesses whose product is the AI agent itself, or where the agent is deeply embedded in a technical service. Cyber liability, by contrast, is generally the wrong policy to reach for: it is built around data breaches and network failures, not the consequences of an agent giving a customer the wrong answer.

The purpose-built category is standalone AI liability insurance. HSB, a Munich Re subsidiary, launched AI Liability Insurance for small and medium businesses in March 2026, distributed through partner carriers. Armilla, a Lloyd's coverholder underwritten by Chaucer, offers limits up to 25 million US dollars per organisation as of January 2026. Testudo, backed by Apollo, Atrium, and QBE, targets mid-to-large enterprises deploying generative AI with limits up to 9.25 million US dollars. None of these three is an employers liability product. All three are the direct answer to the question an SME operator is actually trying to ask when they ask about employers liability: what covers me if my AI agent gets it wrong.

Section 5: What the Air Canada case teaches about where liability actually sits

The British Columbia Civil Resolution Tribunal's decision in Moffatt v. Air Canada, issued 14 February 2024, is the clearest illustration of where AI agent liability actually lands, and it is instructive precisely because it has nothing to do with employment law. A customer asked Air Canada's chatbot about bereavement fares. The chatbot gave incorrect information. Air Canada argued it could not be held responsible because the chatbot was, in its framing, a separate entity from the airline. The tribunal rejected that argument outright, holding that Air Canada was responsible for information provided by its own agent in the same way it would be responsible for information from a human representative.

The result confirms the operator, not any employment relationship, carries the liability. The tribunal did not need to decide whether the chatbot was an employee, because operator liability for an AI agent's outputs does not depend on the agent's employment status. It depends on the agent being the operator's own system, deployed by the operator, acting on the operator's behalf. That is a business liability question, answered by professional liability, Tech E&O, or standalone AI liability policies, not by employers liability, which only ever concerns the employer's duty of care toward its own human staff.

Section 6: A short checklist before you assume you are covered

Step 1. List what your AI agent actually does: who it talks to, what actions it can take, and what data it touches.

Step 2. Ask your broker, in writing, whether your current professional liability, Tech E&O, or cyber policy responds to a claim arising from that agent's actions. Do not accept a verbal assurance.

Step 3. Confirm explicitly that your employers liability policy has not been assumed by anyone in your business to cover the agent by default. It has not, but it is worth a documented confirmation for your own file.

Step 4. If a gap exists, get indications from HSB, Armilla, or Testudo, and start building the governance evidence file a specialist AI underwriter will ask for. The Agent Certified self-assessment at agentcertified.eu is structured around the seven evidence categories underwriters currently request.

Step 5. Review the position annually. AI liability coverage is a market that is still forming, and the gap between what you assume is covered and what is actually written down tends to widen, not narrow, if it is left unchecked.

Section 7: Frequently asked questions

Does employers liability insurance cover my AI agent?

No. Employers liability insurance is a statutory policy that covers bodily injury or illness suffered by a human employee in the course of their employment. An AI agent is not an employee under any current legal framework, cannot suffer bodily injury, and is not covered by an employers liability policy regardless of how central it is to your operations. If your AI agent causes harm to a third party, the relevant policies are professional liability, technology errors and omissions, or a standalone AI liability product, not employers liability.

What did ElevenLabs actually insure in February 2026?

ElevenLabs became the first company to hold an AIUC-1 backed insurance policy for its AI voice agents in February 2026. Coverage was described as treating the agents like any other employee, meaning the company carries a policy that responds to losses the agents cause, in a similar structural role to how employment-related liability cover protects a business from the acts of its staff. The policy itself is an AI liability product underwritten against the AIUC-1 certification standard, not an employers liability policy, and it does not classify the agents as employees for legal purposes.

Which insurance policy actually responds when an AI agent makes a mistake?

For most SMEs, the closest fit among traditional policies is professional liability or errors and omissions, and for technology-heavy businesses, Technology Errors and Omissions. Neither was written with autonomous agents in mind, so coverage is often ambiguous. The purpose-built alternative is standalone AI liability insurance from carriers including HSB, Armilla, and Testudo, or AIUC-1-backed coverage of the kind ElevenLabs holds. None of these are employers liability products.

Can an AI agent legally be treated as an employee for insurance purposes?

No jurisdiction currently recognises an AI agent as a legal employee. Employers liability insurance is built around employment law concepts including bodily injury, occupational illness, and vicarious liability for the acts of a person under a contract of employment. An AI agent has none of these attributes. Framing an AI agent as insured like an employee is a useful way to describe the business intent behind a coverage decision, not a legal reclassification of the agent's status.

What should an SME do before assuming its AI agent has any insurance cover?

Catalogue what the agent does and who it interacts with, request written confirmation from your broker on how your current professional liability, cyber, and Tech E&O policies treat AI-generated outputs, and check specifically that employers liability has not been assumed to cover the agent by default. If gaps are found, approach specialist AI liability carriers or begin a certification process such as Agent Certified to build the evidence file underwriters will ask for.

Section 8: Related reading

On this site:

• Does my business insurance cover AI errors? The 2026 policy-by-policy guide. The full policy-type breakdown this article draws from.
• The Air Canada chatbot case: what SME operators should learn. The full story behind Moffatt v. Air Canada.

Across the network:

• agentinsured.eu. The coverage platform tracking AI liability products entering the European market, including AIUC-1 backed policies.
• agentcertified.eu. The Agent Certified methodology underwriters are beginning to use as evidence before quoting AI agent coverage.

References

1. UK Parliament. Employers' Liability (Compulsory Insurance) Act 1969. Statutory minimum cover requirement of GBP 5 million.
2. AI Underwriting Company (AIUC). AIUC-1 certification and insurance standard, 2025. Reported first policyholder: ElevenLabs, February 2026.
3. HSB (Hartford Steam Boiler, a Munich Re company). "HSB Introduces AI Liability Insurance for Small Businesses." 18 March 2026.
4. Armilla. Affirmative AI liability insurance with Lloyd's underwriter Chaucer, limits expanded to USD 25 million, January 2026.
5. Testudo. AI liability capacity of USD 9.25 million per insured via Apollo, Atrium, and QBE, March 2026.
6. British Columbia Civil Resolution Tribunal. Moffatt v. Air Canada. Decision issued 14 February 2024.