Everything is moving and changing so fast in almost all aspects of life. Accompanied by so much information it’s difficult to keep up with what you’re supposed to be thinking about, let alone make sense of it. It’s happening to everything and everybody, which means this is being experienced by your clients, your suppliers and your peers in pretty much in the same way as you. So not only do you need to worry about yourself, but you need to worry about them, as they will worry about you.
It can be distracting and disturbing if you let it. Those who don’t let it will survive and thrive as they always have.
Marcus Aurelius, the Roman emperor and Stoic philosopher, said: “Never let the future disturb you. You will meet it, if you have to, with the same weapons of reason which today arm you against the present.”
Armed with the ancient laws of logic, we have since discovered the laws of nature and constructed the rule of law. Applying these structures to evidence enables change without change: the same logic applied to new information.
When it comes to AI, it’s apparent from recent speeches and publications that the FCA is taking this approach, reminding us it operates through overarching, outcomes-focused principles that apply universally, regardless of the subject.
Listening to the detail, there seem to be some logical questions we should ask ourselves. What need are we trying to meet with the use of AI? Who is accountable for it and its outcome? What will we do with the output? How can we assure ourselves that it will work and keep working? What do we do if it goes wrong?
Who do we trust as a partner and why? What’s their track record? What’s in it for them? What are their principles? Will they still be around? What’s the evidence?
Evidence for your due diligence and to meet the regulations is one thing, but the question of AI and your relationships with your clients and suppliers goes beyond that – especially if things go wrong. The Financial Ombudsman Service has reported an increase in complaints, alongside the growing use of AI tools to help consumers make them. As we know, bigger cases also go to the courts, so what does the rule of law say about evidence in a world full of audio and video recordings and AI?
- Recording a conversation you are party to isn’t interception under the Regulation of Investigatory Powers Act 2000. GDPR applies to personal data in recordings and transcripts.
- Jones v University of Warwick [2003] ruled that covertly obtained recordings were admitted in a civil case.
- While R v Shephard [1993] and later authority mean courts presume computers work properly unless there’s evidence otherwise, the Horizon post office scandal and subsequent findings show computer infallibility can’t be taken as read when a system’s reliability is genuinely challenged.
- Kogan v Martin [2019] shows human witness evidence still matters even if there are contemporaneous documents, especially on meaning and intention.
So, have AI transcripts or AI use been tested in court?
- Clarke v Guardian News & Media [2025]: Otter.ai transcripts were treated as ordinary disclosable documents. There was no duty to correct them before disclosure, and they were not privileged because the underlying conversations weren’t.
- Employment Tribunal (Rollo): a ChatGPT discussion was held not to be expert evidence and not reliable.
- Godwin v Godwin [2026]: AI-assisted witness statements were approached cautiously because the court couldn’t be sure they were in the witnesses’ own words
Evidence is categorised into types, but there is no definitive ruling in the UK for AI chats, AI transcripts or AI-generated evidence. Courts are treating them case by case under existing rules, but there are several open reviews:
- On 21 January 2025, the Ministry of Justice called for evidence on software-generated evidence. Although the focus is on criminal law, any change is likely to influence how civil courts approach the same question.
- In the same month, the Civil Justice Council began looking at AI use in preparing court documents. The final report is expected later this year.
In the meantime, AI transcripts seem to be treated as ‘derived copies’, weighed by reliability, but with the audio prevailing. Yet with voice copying software prevalent, it’s only a matter of time before this is challenged. Courts care about authenticity, integrity and provenance, and weigh any gaps rather than automatically excluding evidence.
It seems inevitable to me that clients will record the advice given to them and produce their own transcripts at the time, or when they decide to make a complaint. Complaining is easy with AI and that will be placed against the evidence you present, such as an AI created suitability letter. We won’t see this en masse until there are large enough losses to be worth claiming for.
This increase in evidential sources could mean that the old adage of ‘If it wasn’t written down, it didn’t happen’ needs to be revised to, ‘If it didn’t happen, don’t write it down’. A consistent, structured system of frequent and validated, conduct-driven communications means that there should be no conflicting evidence. Or in simple terms, act in your clients’ best interests and tell the truth.
Applying these logical structures, including the FCA principles and rule of law, when considering the use of AI by you, your suppliers and clients will serve you well both today and in the future. And relying on others who do the same will alleviate your worries.
Chris Jones is financial services director at Dynamic Planner
