AI gets things wrong, and you’re still responsible

AI is brilliant at sounding right. That’s the problem. It’ll happily give you a wrong date or a made up figure, and it’ll do it in perfect, confident English.

If that goes out under your name, it’s yours to fix. Your client won’t care that a tool wrote it. So read everything before you send it. It takes a couple of minutes and can save you a lot of embarrassment.

It gets a bit more serious if you’re using AI to help make decisions about people, like who gets shortlisted for a job or whose application gets approved. There are extra rules here. They were loosened recently, but you still need to tell people it’s happening and give them a way to challenge the result. A real person also needs to be able to review it.

And if AI is part of your service, just say so. Most people don’t mind at all. What they mind is finding out later. A quick mention in your privacy notice or terms is a good place to start.

None of this means you should stop using it. I use it most days myself. Just treat it like a keen new helper whose work you’d always check before it goes out.

If you’re not sure where to start, send me a message.