When a manager learns that staff paste client emails into ChatGPT, the first reflex is often a ban. It feels safe and simple. In practice, a ban mostly changes one thing: people stop talking about it.
Why bans rarely work
- The tool is one click away. On a phone or a personal laptop, nobody can really enforce a ban.
- The benefit is real. People who save an hour a day will keep looking for that hour.
- You lose visibility. Use moves to personal accounts, where you cannot check the settings or see what was shared.
- You lose the gains. The time saved stays individual and invisible, instead of becoming a team improvement.
A ban can make sense for a specific tool or a specific type of information. As a general answer to AI, it usually backfires.
What works better: a simple framework
- Give people a good tool. A company account with checked settings removes the main reason to use a personal one.
- Draw clear lines. What is allowed (drafting, summarizing, translating), what never goes in (client identities, contracts, health or financial details), and who decides in a grey zone.
- Keep a person in the loop. Anything that reaches a client, a funder or a court is read by a person first.
- Explain, then sign. A short team session and a signed acknowledgement turn the rules into a shared habit.
- Review twice a year. Tools change fast; your rules should too.
The goal is not less AI. It is AI used in the open, with the right tools and clear limits, so the whole team benefits.
Want to see where your team stands? Take our 2-minute self-check, read how to find out what your employees really use, or let us set it up with you through our Custom AI Policy.