FCA authorised · FRN 724952 0117 325 0027Quote & buy →
Aipedia: IT consultants and developers
If AI-generated mappings or cleaning rules silently corrupt client data during a migration, the client will usually claim the cost of finding and fixing it from the consultant. Data errors often surface months later, after the old system has gone.
Part of: AI in IT consultants and developers
In short
AI tools can propose field mappings, transformation rules and de-duplication logic far faster than manual analysis. The risk is silent error: records that load successfully but are wrong, merged or truncated. Clients have their own legal duties to keep accurate records, including under UK GDPR, so corrupted data can create regulatory and operational problems for them. BCS clauses 3.1 and 3.3 apply to the consultant's method and supervision. Technology PI usually responds to claims for negligent migration, including the client's restoration costs, subject to the wording, while your own data recovery costs may fall to cyber or other cover.
Consultants use AI to profile source data, infer what each field means, suggest mappings between old and new schemas, write transformation scripts, standardise addresses and names, and identify duplicates. On large legacy migrations, that can save weeks of analysis. It also tempts teams to accept proposed mappings in bulk, particularly where field names are cryptic and documentation is missing.
For BCS members, the BCS Code of Conduct requires due care and diligence (3.1) and responsibility for your own and supervised work (3.3). Clause 3.4 prohibits disclosing confidential information, which matters if client data is passed to an external AI tool for profiling. The AI Cyber Security Code of Practice expects data, models and prompts to be documented (principle 8) and data to be disposed of securely at the end of its use (principle 13), both relevant where AI tools are used on client datasets.
The client usually remains the controller of personal data and is responsible for its accuracy and integrity under UK GDPR. A migration error can therefore become the client's compliance problem as well as an operational one. The ICO's AI guidance hub covers data protection in AI use generally.
Illustrative scenario: a consultancy migrates a housing association's tenant records to a new platform. An AI tool proposes mappings, including one that treats a legacy "status" code as a contact preference. Validation checks only that row counts match. After go-live, vulnerable tenants flagged for priority repairs lose their flags, and duplicates merged by AI combine separate households. Problems emerge over several months, by which time the old system has been decommissioned. The client claims the cost of reconstructing records and handling complaints.
Silent truncation of free-text notes, wrong date formats and currency or unit conversions are other recurring failure points.
Where the client claims that your migration was negligent, technology PI commonly responds to the claim, which may include the client's costs of investigating, restoring and correcting data, subject to the wording, limit and excess. Your own costs of re-running work to put things right, without a claim, may not be covered, and some wordings treat rectification costs differently. If data loss followed a security incident, cyber cover may respond. See PI for data migration specialists and scope of duty in IT claims.
Many migration disputes turn on acceptance. If the client signed off the migrated data after reasonable testing, it may be harder for it to argue later that errors were all the consultant's fault. That only works if acceptance criteria are meaningful. Agree in the statement of work what the client will test, how long it has to raise defects, which reconciliations you will provide and what happens to errors found after sign-off. Where AI tools proposed the mappings, say so, and give the client a readable list of mappings to review. That transparency puts the client in a position to spot business meaning errors that a technical team may miss.
Apex arranges technology PI for IT consultants and data specialists. A broker reviews your cover and how your policy treats data restoration claims from AI-assisted migrations. See also LLM integrations and data leaks, AI in managed services and the IT consultants AI hub.
The client usually remains responsible for its records as controller. The consultant can still be liable to the client in contract or negligence if a migration error was caused by a lack of reasonable care, such as unchecked AI-proposed mappings.
If the client claims those costs from you as a result of negligent work, technology PI commonly responds, subject to the wording and limit. Your own costs of re-doing work without a claim, or of recovering data after a cyber incident, may fall elsewhere.
Only if the contract and data protection arrangements allow it. BCS clause 3.4 prohibits disclosing confidential information, and the client must be content with where the data goes, how long it is kept and whether it is used for training.
Long enough for the client to validate the migrated data in real use. Agree this in the contract. Decommissioning too early removes the easiest route to correct silent errors and can increase the cost of any claim.
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 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 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.
