Whose data is it anyway?
AI, consent and the trust at the heart of the profession
The final episode of the CCAB’s podcast series opens with a disclosure its AI hosts have made all series: the two of them are AI voices, and the podcast was created using AI. As Sam notes, that is worth holding in mind for an episode about who owns data and how it gets used.
Because this episode is all about data: who it belongs to, who consented to what, and what happens to professional trust when AI blurs those lines.
Training your AI on client secrets
The first scenario takes place at a big global firm developing its own large language model to enhance audit, tax and advisory services. To train it, the technology team proposes using years of confidential client data to generate ‘synthetic’ data, a statistically equivalent version that does not contain actual client information. Confidentiality maintained, they argue. Let’s move.
The episode carefully takes that argument apart. Synthetic data is created from actual client data, and the processing step is itself a use of confidential information that clients never consented to. The IESBA Code is explicit: client authorisation is required before confidential information is used in the development of technology. The Financial Reporting Council has clarified the same point. And synthetic data generation is not perfect; there is a small but real risk of actual data leaking through, which for clients in regulated industries may cross a line on its own.
The suggested answer is not to abandon innovation but to do it properly: obtain explicit client authorisation, develop a genuine consent framework with opt-ins and opt-outs, and pilot with clients who have actually agreed. The issue is doing it without asking.
The classroom, the algorithm and the false flag
The second scenario moves to a university accounting department piloting an AI platform that gives feedback on assignments and checks for plagiarism. The rationale is sound, but the pilot turns up serious problems. The feedback on complex case studies is often generic and occasionally wrong. The plagiarism checker starts flagging international students whose writing patterns differ from native English conventions, putting innocent students under investigation. And the vendor’s vague terms of service leave open whether student work is being used to train a commercial AI, a GDPR compliance problem as well as an ethical one.
Perhaps the most important question the episode asks is what this models for the next generation of Chartered Accountants. If we teach students that it is acceptable to delegate judgement to an AI and not look too closely at how it works, we have given them a very dangerous professional habit.
The line that closes the series
The accounting profession requires an inquiring mind, the exercise of professional judgement and ethical decision making with clear accountability. Those things cannot be automated. They have to be taught, practised and earned.
Across three episodes and six case studies, Ethics in the Age of AI has covered AI in strategic decisions, audit, financial reporting, data privacy and education. The question it leaves the profession with is the one worth keeping: not just ‘can we do this?’ but ‘should we?’
Listen to this episode of Ethics in the Age of AI below or subscribe on Apple Podcasts, Spotify, Amazon Music or your favourite podcast library, and read the complete CCAB case studies with all the ethical considerations and recommended courses of action at ccab.org.uk