T-001 · Data boundaries
Don't paste customer data into public AI tools
Before you paste it into an AI tool, ask: would I be comfortable if this left my company? Customer details stay out of tools you haven't approved.
Read the 30-second version →AI Safety Tips
Short, practical guidance on using AI at work. Every tip has a 30-second version and a line that fits on a label.
T-002 · Data boundaries
Keep contracts, pricing and HR notes out of unapproved tools
Some information is too sensitive for a quick AI shortcut. Keep contracts, pricing and personnel notes out of tools your team has not approved.
Read the 30-second version →T-003 · Data boundaries
Ask "who can see this?" before you hit send
Every prompt goes somewhere. Before you send one, ask who can see it, how long it is kept, and whether you are allowed to share it.
Read the 30-second version →T-004 · Verify outputs
Confident AI can still be wrong
AI sounds sure even when it's guessing. Check names, numbers and sources before you use them.
Read the 30-second version →T-005 · Verify outputs
Check every name, number and source
Names, numbers and sources are where AI slips most. Check each one against the original before it goes into a document, email or decision.
Read the 30-second version →T-006 · Verify outputs
If it cites a link, open the link
AI can invent sources that look real. If an answer cites a link, article or study, open it and confirm it says what the AI claims.
Read the 30-second version →T-007 · Permissions
Read what an AI tool wants access to before you say yes
Before you connect an AI tool to your email, files or calendar, read what it asks to see and do. Say yes only to what the job needs.
Read the 30-second version →T-008 · Permissions
Connect one account at a time
Start with a single low-risk account or folder. Connect more only after you know how the tool behaves.
Read the 30-second version →T-009 · Human review
A person reads it before it is sent or published
AI can draft. A person decides. Anything sent to a customer or published under your name gets read by a human first.
Read the 30-second version →T-010 · Human review
Don't let AI make the final call on people or money
Hiring, firing, pricing and payments deserve human judgment. Use AI to prepare. Let a person decide.
Read the 30-second version →T-011 · Getting started
Start with low-risk tasks
Begin with drafts, brainstorms and summaries of public information. Save sensitive work for after you know the tool.
Read the 30-second version →T-012 · Getting started
Write a one-page AI rule for your team
A short, plain rule beats no rule. Say which tools are approved, what never goes in, and who reviews what.
Read the 30-second version →Want these on your bottle? See the Custom Water page or the Label Kit.
More resources
Coming next to the AI Safety Hub.
AI Safety Checklist
A one-page checklist for using AI at work, built from the tip library.
AI Policy Starter Kit
Plain-language starting points for a small team’s AI rules.
AI Readiness Check
A short self-assessment to see where your team stands.