Using AI to write your ombudsman complaint, and why it can backfire

The ombudsman is picking up the phone more often. The reason is not a new service promise. It is the chatbot that wrote the complaint.

Chief ombudsman James Dipple-Johnstone told the Financial Times that in the most striking case he had seen, 37 of 39 references in one consumer’s submission were wrong, as Insurance Business reported on 27 October. The service’s answer is more telephone contact, to get to the heart of the dispute.

This is not a fringe problem. In its response to the Mills Review in February, the ombudsman estimated from a small sample that AI may have contributed to up to a third, 35 per cent, of responses to its initial assessments. It added that heavy use had pushed some parties into more entrenched positions.

It lands as the queue grows. The ombudsman’s quarterly data shows 4,400 car and motorcycle insurance complaints between July and September, up from 4,100 the quarter before and 3,200 a year earlier, mostly about what the claim was worth. The service names the growing use of AI tools as one possible factor in the wider rise.

Here is the detail worth sitting with. A dispute about an insurance claim is won on facts: the date, what was damaged, what the insurer said and when, and what you lost. An invented ruling adds nothing to any of that. It buries it, and someone then has to spend time proving the ruling does not exist.

The ombudsman’s own guidance on using AI is sensible. Use it to organise your notes or put them clearly, check every line, keep health and banking details out of it, and remember the service is free, so you do not need a lawyer or a claims firm.

If a tool tells you a case says something, ask it for the link. If it cannot give you one, delete the sentence.

Your own words, backed by your own evidence, are still the strongest complaint you can make.