Purpose
Morrell Middleton uses artificial intelligence (AI) tools to work more efficiently. This policy explains how our team uses them responsibly: protecting client and company information, using AI ethically, and managing the risks it brings.
Scope
This policy applies to all our staff, contractors and anyone else with access to our systems, data or AI tools.
What we mean by AI
Any software that uses machine learning, language processing or automation to analyse data, generate content, make predictions or support decisions. This includes chatbots, text generators, analytics tools and AI features built into other software.
Protecting your information
Staff must never enter confidential client or company information into an AI tool unless that tool has been approved by a director for that purpose. Free public AI tools may only be used for general research, and never with client data. Approved business tools must be used while signed in to our company accounts, so that our data protection settings apply.
Approved AI tools
The AI tools approved for use by our team are: Claude (by Anthropic), used for research, drafting and efficiency tasks through our company account; Microsoft Copilot, used within our Microsoft 365 company account so our data protection settings apply (the free public version must not be used for work); and Lovable, used to build and maintain our websites, which contain no client data. Any other AI tool may only be used for general queries, and never with client or company information.
AI in other software
Many platforms we use, such as accounting software, now include AI features. These are only used in ways that meet our data protection, confidentiality and security obligations under UK GDPR. New AI-enabled tools must be reviewed and approved by a director before they are used with client data.
The risks of AI
AI can expose data if it is not properly secured, and it can produce answers that are biased, out of date, inaccurate or simply made up. We take these risks seriously.
Human review
Our staff check anything produced with AI against trusted sources before using it. Nothing AI-assisted is sent to a client without review by a qualified member of our team, and the person who produces it is responsible for its content as if they had written it themselves. AI never replaces professional judgement on your affairs.
Reporting concerns
Staff must report any suspected misuse of AI, or any accidental sharing of data with an unapproved tool, to a director straight away.
Questions
If you would like to know more about how we use AI, or would prefer that AI tools are not used on your work, please contact us at reception@gravitax.co.uk or on 01904 691141.