Aipedia: Occupational health providers

AI mental health chatbots in employee support

Aipedia, by the Apex broking team · Last reviewed 2026-10-27

OH providers that offer AI wellbeing or CBT chatbots to workers need clear escalation for risk and clarity about who is clinically responsible. A chatbot that misses signs of self-harm can cause serious harm and leave the OH provider, the employer and the tech supplier disputing liability.

Part of: AI in occupational health providers

In short

AI chatbots offering wellbeing support, CBT-style exercises or signposting may be bundled into OH and employee assistance contracts. They are available at any hour but cannot assess risk as a clinician can. If a worker discloses suicidal thoughts and the chatbot fails to escalate, the OH provider may face claims and regulatory scrutiny of the clinicians who oversee the service. The NMC, HCPC and GMC codes apply to registrants who supervise or recommend the tool, and health data handling must meet ICO guidance. Whether OH malpractice cover or the supplier's technology cover responds depends on the contracts and wordings.

How chatbots are offered through OH

Wellbeing chatbots may be supplied by a technology company and offered to workers through an OH or employee assistance programme contract. They may provide mood tracking, guided CBT-style exercises, sleep and stress content, and signposting to human services. Some can book an OH or counselling appointment. Employers may like them because they are cheap to scale and available outside office hours. Workers may use them when they would not speak to a person.

That last point is both the benefit and the risk. A worker in distress may disclose more to a chatbot than to their manager or GP, so the moment of disclosure is exactly when the tool has to respond safely and hand over to a person.

The duties that apply

Registrant accountability: practitioner psychologists and other HCPC registrants, OH nurses and occupational physicians remain accountable under their codes for services they design, supervise or recommend. The GMC's AI learning resource says doctors apply Good medical practice to AI and report incidents involving software (para 73).
Confidentiality: NMC Code 5.1 and 5.2 and HCPC standards 5.1 and 5.2 require privacy and telling people how their information is used. Workers need to know whether anything they type reaches their employer.
Health data: the ICO workers' health guidance expects detailed health information to stay with health professionals and requires an Article 9 condition for processing.
Regulator direction: the September 2026 joint statement commits regulators to UK-wide AI principles, which may address supervision of AI tools.

What can go wrong

Illustrative scenario: a worker uses a wellbeing chatbot offered through their employer's OH contract late at night and types that they cannot see a way forward. The chatbot responds with breathing exercises and a link to articles, without recognising the risk or giving crisis contacts. The worker later attempts self-harm. The family asks the employer what checks were made, the employer points to the OH provider, and the OH provider points to the chatbot supplier's terms.

Other risks include chatbots giving advice that conflicts with the worker's treatment, data shared with the employer in ways workers did not expect, and workers using a chatbot instead of seeking an OH referral they needed. The duty of care question is who designed the escalation pathway and who monitored it.

How PI usually responds

An OH professional indemnity or medical malpractice policy usually covers clinical services delivered by the OH provider and its clinicians, subject to its terms. A chatbot run entirely by a technology supplier may be treated as the supplier's product, and one medical defence organisation's published position excludes autonomous systems and software defects. The supplier's own technology and liability cover may then be the relevant policy. Check the contract chain, indemnities between the parties, and how bodily injury claims are treated. See AI chatbot insurance, insurance for healthtech companies and does OH indemnity cover AI mistakes.

Practical controls

Test the chatbot's risk detection and escalation before offering it.
Make sure every risk disclosure produces crisis contacts and a route to a human.
Name a clinician responsible for clinical oversight and incident review.
Tell workers clearly that the chatbot is not a clinician and what happens to their data.
Agree in contracts who is responsible for content, monitoring and incidents, and require the supplier to hold suitable cover.
Tell your insurer that the service is part of your offering.

How Apex helps

Apex arranges professional indemnity for occupational health providers. A broker reviews whether your PI responds to digital mental health services you offer and how it fits with your supplier's cover; see OH professional indemnity.

Sources

Frequently asked

Is an OH provider responsible for a chatbot it offers?

It may be. If the provider selects, recommends or integrates the chatbot into its service, it can be asked what checks it made and how risk was escalated. Clinicians who supervise the service remain accountable under their professional codes.

Will OH malpractice cover respond to a chatbot claim?

It depends on the wording. OH professional indemnity usually covers clinical services by the provider and its clinicians, but one defence organisation's published position excludes autonomous systems and software defects. The supplier's own cover may be the main policy, so contracts and indemnities matter.

What escalation should a wellbeing chatbot have?

At a minimum, recognition of risk language such as self-harm or suicide, immediate crisis contacts, and a route to a human clinician. A named clinician should review incidents and test the pathway regularly.

Can the employer see what workers tell the chatbot?

It should not see individual health disclosures. The ICO says detailed health information should stay with health professionals, and workers must be told how their data is used. Employers may receive aggregated usage data if this is explained.

Check your indemnity covers how you use AI

Tell us about your practice and the work you do. A broker will review how your current cover treats AI-assisted work, or arrange new cover. No fee for a quote. Or call 0117 325 0027.

Start an indemnity quote Request a callback

Apex 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 about how AI is used in the profession and how insurance may respond. It is not legal, regulatory or insurance advice on your circumstances, and it does not guarantee that cover will be available or on what terms.