A conveyancing client sends an email: “I have checked this with an AI tool. It says the indemnity policy deals with the problem, so please go ahead and exchange.”
The answer may look convincing. It may explain the legal terminology, identify a familiar type of insurance and present a confident conclusion. But did the tool see the complete title, the policy exclusions, the lender’s requirements and the correspondence explaining why the buyer’s solicitor remains concerned?
This is an illustrative scenario, but it exposes a practical risk for conveyancing firms: a client can arrive at a firm decision using information that nobody responsible for the transaction has verified.
Firms need a clear process for responding when chatbot output starts influencing instructions, expectations and decisions.
The client may be asking a different question
Someone buying a home may reasonably use AI to explain unfamiliar terms, organise questions or make a lengthy report easier to approach. Those uses can help a client participate in the transaction.
The difficulty arises when an explanation becomes a recommendation to act.
Consider the difference between asking “What is restrictive covenant indemnity insurance?” and asking “Can I safely buy this house?” The second question depends on the actual documents, the proposed use of the property, the lender and the client’s objectives.
A chatbot may answer the question it receives without identifying the question the client should have asked. The conveyancer needs to establish which assumptions sit behind the resulting instruction.
Identify what the client wants you to do
When AI-generated material arrives, begin with the decision it is intended to support. Is the client asking for an explanation, challenging existing advice, requesting additional work or instructing the firm to proceed despite a warning?
Ask for the relevant output and, where useful, the question and documents supplied to the system. Request only what is reasonably necessary to understand the issue. Clients should not be encouraged to circulate unrelated personal information or an entire chatbot history.
The file should distinguish:
- the client’s factual information;
- the chatbot’s assumptions and conclusions;
- the client’s proposed instruction; and
- the firm’s own advice on the issue.
That distinction prevents an externally generated statement from quietly becoming an accepted fact within the conveyancing file.
Three conveyancing assumptions worth challenging
“The insurance means the defect does not matter”
A client may interpret an insurance recommendation as confirmation that the underlying title problem has disappeared. The conveyancer should explain what the proposed policy actually covers, its exclusions and any outstanding lender or transaction requirements.
The question is whether the proposed arrangement addresses the identified problem for this transaction. A general explanation of indemnity insurance cannot establish that.
“My lender accepted the property before”
Previous mortgage acceptance does not determine what a different lender will accept now. Nor does today’s requirement establish what applied to an earlier purchase.
If the client is questioning the original transaction, identify the relevant lender, date, documents and advice. If the instruction concerns the current sale, establish the present buyer’s lender’s actual objection.
“The summary does not mention any serious restriction”
An AI summary may omit a qualification, an exception or a restriction contained in another document. A short explanation of a lease is not confirmation that the client’s intended letting, alterations or business use is permitted.
Ask what the client intends to do with the property and check the relevant documents. That conversation may reveal an issue that the summary never addressed.
Receiving AI output does not make it the firm’s advice
A firm should acknowledge material without creating the impression that it has approved every statement within it. Where the client requests a review, explain what will be examined, what information is needed and whether additional work or fees must be agreed.
However, scope wording must match the firm’s conduct. If the conveyancer adopts an assertion in a report, relies on it in advice or uses it to justify proceeding, describing the original material as “client supplied” does not replace proper verification.
The SRA’s warning notice on misuse of AI, published on 17 August 2026, emphasises that professional responsibility remains with solicitors when AI is used in delivering legal services. It also identifies risks involving inaccurate information and confidentiality.
For conveyancers, the practical response is to verify material before relying on it and explain the resulting advice in terms the client can understand.
A client’s willingness to proceed may not resolve the problem
A signed instruction recording that the client accepts a risk can be important evidence. It is not a universal permission to proceed.
The firm must still consider its obligations to any lender client, applicable professional duties and whether it can properly carry out the proposed instruction. A borrower’s acceptance cannot supply an approval that must come from the lender.
Likewise, asking the client to sign a broad disclaimer is no substitute for explaining the specific concern and the available options.
Address confidentiality before documents are uploaded
Clients may paste a Report on Title, correspondence or identification documents into an external AI system without considering how that information will be processed.
Firms should explain the risks early. The implications depend on the material, the provider’s terms, the system’s safeguards and the circumstances of disclosure. Uploading information should not be described as automatically making every document public, but it can create confidentiality, privilege and data protection concerns.
Lexsure’s Knowledge Hub includes a directly relevant paragraph on client use of artificial intelligence systems. Its stated purpose includes warning about disclosure risks and clarifying that external AI output has not been reviewed or approved by the firm unless expressly confirmed.
Connect client care wording with the firm’s procedures
Client-facing terms should be supported by an internal process. Staff need to know when to clarify instructions, verify material, involve a supervisor and record the advice given.
Lexsure’s AI and technology partners policy paragraph addresses professional responsibilities around technology use. Firms reviewing their wider controls can also consider the CQS CPMS Risk Management Policy template, adapting it to their actual systems and working practices.
The client-facing clause and internal policy serve different purposes. One explains the relationship to the client; the other tells staff how to manage the risk.
Our article on AI and professional judgment in conveyancing examines the firm’s own use of automated analysis. The separate article on responding to AI-generated conveyancing complaints considers what happens when externally generated material becomes a formal expression of dissatisfaction.
Help clients ask better questions
A client who uses AI is not necessarily rejecting professional advice. They may want a quicker explanation, reassurance about costs or help understanding a difficult document.
Explain why the particular facts change the answer. Identify the decision that remains outstanding and tell the client what is needed to resolve it. Clear, accessible advice gives the client something more useful than a contest between two apparently confident conclusions.
Do your client care terms address clients using AI?
Review whether your terms explain confidentiality risks, the status of externally generated output and how any requested review will be agreed.
Lexsure’s CLIENTCAREmonitor Terms and Conditions Notification Service provides notifications of wording for firms to consider and adapt as client behaviour and working practices change.