Aipedia: Insolvency practitioners

AI chatbots for creditor and debtor queries

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

AI chatbots can answer routine creditor and debtor questions, but wrong answers about claims or dividends are the office holder's responsibility. Confidentiality failures add a data protection risk that may fall to cyber rather than PI.

Part of: AI in Insolvency practitioners

In short

Creditor portals and chatbots can handle routine queries about claims and procedures. The risks are confident wrong answers on dividends or deadlines that creditors rely on, and disclosure of one party's data to another. The Insolvency Code of Ethics requires the IP to decide whether output is appropriate (R321.1) and to protect confidentiality (R104.3 to R104.6). PI usually responds to negligence claims from creditors, subject to the wording, while data breach costs usually sit with cyber cover. Limit chatbots to process questions.

How chatbots and portals are used

Every insolvency generates questions: how to submit a claim, when a dividend might be paid, what happens to a debtor's car, why a payment was stopped. Creditor portals and chatbots promise to answer these at any hour without tying up case staff. R3's 2026 report lists knowledge chatbots among current uses, mainly internal, and some firms may extend the idea to external queries.

A chatbot answering from general knowledge is very different from one answering from a specific case file. Both carry risk, in different ways. A general chatbot may give a confident but wrong account of insolvency law or procedure. A case-linked chatbot may give an accurate answer to the wrong person, or an answer based on data that was right last quarter but is wrong now. Creditors and debtors are unlikely to distinguish between a chatbot answer and a statement from the office holder.

The duties that apply

Answers given through a firm's chatbot are communications from the office holder. Under the Insolvency Code of Ethics:

R321.1: the IP must determine whether the chatbot's output is appropriate for the intended purpose, which for external users means answering accurately without supervision.
R104.3, R104.4 and R104.6: confidentiality. A chatbot linked to case data must not disclose one creditor's or debtor's information to another.
R320.5: where the chatbot is a third-party service, review the arrangement periodically.
R130.2: keep records of what was said, so the firm can show what a creditor was told.

UK GDPR also applies to personal data handled by the chatbot and its provider.

What can go wrong

Illustrative scenario: a creditor asks a firm's portal chatbot whether there will be a dividend in a liquidation. The chatbot, drawing on an outdated estimate, says a dividend of a stated percentage is expected later in the year. The creditor writes the debt down by less than it otherwise would and decides not to pursue a guarantor. The dividend never materialises, and the creditor complains and seeks compensation.

Other failure points:

wrong guidance on claim deadlines or how to prove a debt;
confident answers on legal questions the chatbot should have referred to a person, a pattern known as AI hallucination;
disclosure of another party's information through a poorly scoped connection to case data;
a security weakness that exposes portal data.

Whether PI or cyber responds

A creditor who relies on a wrong answer and suffers loss may bring a negligence or misrepresentation claim. PI usually responds to that type of claim, subject to the wording, though some policies may limit cover for statements made outside a formal engagement or through automated channels. R3 found some practitioners unsure whether PI covers AI errors at all.

A data breach through the portal is mainly a cyber risk: incident response, notification and system costs usually fall under cyber cover, while third-party claims for breach of confidence may fall under PI, cyber or both depending on the wordings. See where cyber and PI cover begin and end and combining cyber and PI.

Practical controls

Limit the chatbot to process questions, and route anything about a specific claim, dividend or legal right to a person.
Show clear wording that answers are general and that formal information is in statutory reports.
Make sure any case-data connection enforces strict access controls per user.
Keep logs of conversations and review a sample regularly.
Check that the provider does not reuse data; see debtor and creditor data in AI tools.
Keep chatbot content consistent with progress reports and the claims register.

How Apex helps

Apex arranges PI for insolvency practitioners, and a broker can review how PI and cyber cover respond to creditor portals and chatbots. See the PI guide for IPs and the insolvency practitioners hub.

Sources

Frequently asked

Is the IP responsible for what a chatbot tells creditors?

In practice, yes. A chatbot run by the firm speaks for the office holder, and R321.1 requires the IP to decide whether its output is appropriate for the purpose. Limiting it to general process questions reduces the risk.

Should a chatbot give dividend estimates?

It is safer not to. Dividend prospects change and depend on case-specific facts, so creditors should be pointed to the latest statutory report or a member of the team. A wrong estimate that a creditor relies on could lead to a claim.

Does PI or cyber cover a chatbot data leak?

Usually cyber cover handles incident response and notification costs. Third-party claims for breach of confidence may fall under PI, cyber or both, depending on the wordings, so check how your policies fit together.

Check your PI covers how you use AI

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

Start a PI 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.