Guide
How to use AI safely at work
Using AI safely at work is less about the technology and more about a handful of habits. These six cover most of the risk for most people.
Six habits
- Use the tools your employer has approved, signed in with your work account.
- Before you paste, decide whether the information is public, redactable, or off limits. Redact names and identifiers when the task doesn't need them.
- Treat every answer as a draft. Check facts, figures, names and quotes against the source before anything leaves your hands.
- Keep a person in charge of decisions about people: hiring, discipline, grading, eligibility.
- Say when something was substantially drafted with AI, especially if a customer would reasonably expect to know.
- If something went in that shouldn't have, tell whoever looks after data protection straight away. Speed matters more than blame.
Why checking matters so much
AI tools produce the most likely-sounding answer, not a looked-up fact. That makes them good at drafting and summarising and unreliable on specifics: names, dates, figures, citations and anything recent. The failure is quiet, because a wrong answer reads exactly as confidently as a right one.
A short routine fixes most of it: check the specifics, ask where each claim came from, and get a second opinion on anything that matters. The free Prove It course teaches that routine.
For managers
If your team is using AI without agreed rules, the fastest fix is a one-page set of team rules that people helped write. Start from a draft, talk it through, and review it when your tools change.