AI Use Policy

Last updated: 30 October 2026 (Version 1.0). Next review: October 2027.

Purpose/Aim/Scope

This policy sets out how Best Ever Medical Devices (BEMD) uses artificial intelligence (AI) in its business, including on its websites and social media platforms. It makes sure AI is used safely, honestly and lawfully, that every piece of AI-assisted work is checked by a person before it is used, and that AI never has access to personal or financial information.

This policy should be read alongside the BEMD Data Protection Policy. It is based upon:

  • The Data Protection (Bailiwick of Guernsey) Law, 2017
  • Guidance on artificial intelligence issued by the Guernsey Office of the Data Protection Authority (ODPA)

This policy applies to all employees (permanent, contractors, management, temporary staff and volunteers) of BEMD, and to anyone who uses AI tools on BEMD’s behalf.

Definitions

Artificial Intelligence (AI): Computer software that can carry out tasks that normally need human thinking, such as writing text, answering questions, editing images or following instructions.

Claude: The AI assistant made by Anthropic, which BEMD uses as its approved AI tool. This includes Claude when it is used through a web browser, the Claude desktop app, Cowork, and the Claude in Chrome browser extension.

Personal Identifiable Information: Any information that relates to a living person who can be identified from it, directly or indirectly. Examples include names, addresses, email addresses, telephone numbers, order details, photographs of identifiable people, and the content of private messages.

Personal Financial Information: Any information about a person’s finances or payments, including card details, bank details, payment records and invoices.

Task: A specific, clearly described piece of work that the Director asks Claude to carry out.

Approved AI Tool

BEMD uses Claude, made by Anthropic, as its approved AI tool. No other AI tool may be used for BEMD work without the written approval of the Director.

Claude is used only as a tool. It works only on Tasks it has been given, and it does not make decisions on behalf of BEMD.

How AI is Used

Claude may be instructed to carry out Tasks such as:

  • Drafting and editing blogs, website text, social media posts and other marketing material
  • Preparing and improving product images, diagrams and graphics
  • Making changes to the content of BEMD’s websites when instructed
  • Researching topics and gathering publicly available information
  • Drafting and updating company policies and other business documents

Claude must not be used to:

  • Make decisions about individuals, including customers, staff or suppliers
  • Reply to customers, or send any message on BEMD’s behalf, without the Director first approving the exact wording
  • Give veterinary, medical, legal or financial advice
  • Make purchases, payments or any other financial transactions
  • Create false or misleading content, such as invented reviews or testimonials

Checking and Approving AI Work

Every piece of work Claude carries out is checked by a person before it is used. In particular:

  • Claude reports back to the Director when each Task is finished
  • The Director checks all of Claude’s work for accuracy, quality and suitability before it is commissioned, used or published
  • Nothing is published on BEMD’s websites or social media, or sent to anyone outside BEMD, until the Director has approved it
  • Facts, figures, references and product claims are checked, and all claims about the Horse Tendon Pad must stay within what BEMD can support
  • BEMD remains fully responsible for all content it publishes, whether or not AI helped to create it

Personal and Financial Information

Claude is not permitted to access, view, collect, store or process any Personal Identifiable Information or Personal Financial Information. This applies to BEMD’s files, websites and social media platforms. To make sure of this:

  • Claude is never given customer lists, order details, enquiries, staff records or any other files that contain personal information
  • Claude only works on BEMD’s websites through its own limited Editor login. This login cannot see orders, customers, payments, users, plugins or website settings
  • Claude must not open the comments or product review areas of BEMD’s websites, as these contain the names and email addresses of the people who left them
  • Claude must not read private messages, comments or follower details on BEMD’s social media platforms
  • Claude never has access to payment systems, bank accounts or card details
  • Passwords and log-in details are entered by the Director, never by Claude
  • Photographs are checked before use, and identifiable people are only shown with their permission

If Personal Information is Seen by Mistake

If Claude comes across Personal Identifiable Information or Personal Financial Information by mistake, it must stop the Task, must not record or use the information, and must tell the Director straight away. The Director will treat this as a possible data breach and follow the Reporting a Data Breach procedure in the BEMD Data Protection Policy.

Account and Security

BEMD’s Claude account is held by the Director. The Director will:

  • Keep the account secure and not share log-in details
  • Review Claude’s privacy and data settings at least once a year, and choose settings that stop BEMD’s work being used to train AI models wherever that option is available
  • Make sure Claude’s memory features are not used to store personal or financial information
  • Only connect Claude to BEMD’s websites, files or social media for as long as is needed to complete a Task
  • Use Claude for website work only in a separate Chrome browser profile, set up for that purpose, which is logged in with Claude’s limited Editor login
  • Keep the password for Claude’s Editor login secure, enter it personally, and never save it in the browser or share it with Claude

Honesty and Transparency

BEMD will be open about its use of AI. AI-assisted content must be accurate and must not mislead. Images may be improved, for example by adjusting brightness or removing third-party brand names, but must not be changed in a way that misrepresents the product or what it does. BEMD will not use AI to create reviews, testimonials or endorsements, or to imitate real people.

Copyright

AI-assisted content must not copy other people’s work. BEMD will only use its own photographs and videos, images it has permission to use, or original material. Quotes and sources must be properly credited.

Roles and Responsibilities

Director

The Director, Jim Marshall, is responsible for this policy. The Director gives Claude its Tasks, checks and approves all of its work, and makes sure this policy is followed.

All Staff

All staff must follow this policy, use only the approved AI tool, and report any concern about the use of AI to the Director.

Breaches of this Policy

Any breach of this policy must be reported to the Director straight away. Any breach involving personal information will be handled under the BEMD Data Protection Policy. Any employee who breaches this policy may face disciplinary action.

Monitoring and Review

The Director will review this policy at least once a year, and sooner if BEMD changes the AI tools it uses, the way it uses them, or if the law or ODPA guidance changes.

References

ODPA Artificial Intelligence Guidance: www.odpa.gg/guidance/artificial-intelligence

ODPA Ten-Step Practical AI Guidance: www.odpa.gg/guidance/ten-step-ai-guidance

Anthropic Privacy Policy: www.anthropic.com/legal/privacy