HomeResourcesAI usage policy
Responsible use of AI

AI usage policy

A few simple principles for using AI at Proteor in a way that's helpful, safe and true to our “people first” promise. When in doubt, ask — that's always the right move.

Our principles

Five guideposts to get it right

People stay in charge

AI assists, it doesn't replace your judgement. A person is always responsible for the final decision — especially anything that affects a patient.

Protect the data

Only put in information you're allowed to share, and prefer approved tools that offer enterprise data protection.

Verify before you use

AI can be confidently wrong. Review and cross-check facts, figures, names and code before relying on them.

Be transparent

Be open when meaningful content was produced with AI's help, and ask for agreement before recording a meeting.

Stay within approved tools

Use the tools recommended on this hub. For anything else, go through “Request AI Tool” before using it for work.

Respect sector rules

Our medical work carries strict rules (patient confidentiality, quality, regulatory). AI doesn't override them — it works within them.

In practice

Do's and don'ts

Do

  • Sign in with your Proteor account so enterprise protections apply.
  • Review and correct any output before you send, publish or ship it.
  • Anonymise or generalise examples whenever you can.
  • Ask participants' agreement before recording a meeting.
  • Ask for help when unsure — the team is here for exactly that.

Don't

  • Enter patient, medical or personally identifying data.
  • Paste secrets, passwords, API keys or sensitive proprietary code.
  • Use unapproved consumer tools for Proteor data.
  • Make a clinical or regulatory decision on an AI output alone.
  • Share generated content you haven't checked and taken ownership of.

What never to put into an AI tool

Whatever the tool, keep this information out of prompts unless you have explicit approval and a tool cleared for it:

  • Health or identifying data about patients or device users.
  • HR or personal information about colleagues.
  • Trade secrets, unpublished designs, sensitive regulatory data.
  • Credentials, passwords, API keys, confidential source code.
  • Anything covered by a confidentiality agreement or NDA.

A question?

When in doubt, ask

Not sure whether something's allowed, or want to suggest a new tool? Reach the team that looks after AI at Proteor.

Request AI Access Request AI Tool