Handling Information

Using AI Tools Safely

Foundational

AI tools like chatbots and assistants are genuinely useful — for drafting, summarising, and getting unstuck. But anything you type into one may leave the company and be stored or used elsewhere. The simple rule: use approved AI tools, never put customer or confidential data into them, and always check what they give back.

When you paste text into a public AI tool, you're sending it to an outside company's systems. You can't assume it stays private or gets deleted. So while AI is a great helper for general, non-sensitive work, it must never become a place where customer data, personal information, secrets, or confidential company material ends up.

AI also sounds confident even when it's wrong — it can invent facts, figures, and details. Treat its output as a useful draft to check, not a trustworthy answer to rely on, especially for anything that matters.

Use AI well

DoUse approved AI tools for general, non-sensitive tasks — drafting wording, summarising public information, explaining concepts, brainstorming.
DoCheck and verify what AI gives you before using or sharing it — it can be confidently wrong, and you're responsible for the result.
DoKeep your prompts free of anything you wouldn't be happy to send to an outside company.
ConsiderAsking your manager or security if you're unsure whether a tool is approved or whether something is safe to put in.

The hard line

AlwaysTreat anything you type into an AI tool as if it could leave the company — so keep it strictly non-sensitive.
NeverPut customer data, personal information, identity documents, passwords, secrets, or confidential company material into a public or unapproved AI tool.
NeverRely on an AI's answer for an important decision without checking it against a trusted source.

Ask yourself

AskWould I be comfortable if everything in this prompt left the company? If not, don't paste it.
AskIs there any customer or personal data, or anything confidential, in what I'm about to enter?
AskHave I checked whether the AI's answer is actually correct?

Why it matters: Pasting customer or confidential data into a public AI tool can be a data breach in itself — the information has left our control. Used sensibly, with no sensitive data and a check on its output, AI is a great productivity boost; used carelessly, it's a fast way to leak exactly the data we're trusted to protect.

Tips

Tip Staying in control Claude rules that keep you in control
Tip Webinars Anthropic