It feels quite cosy, doesn't it. Just you, your laptop, and your AI having a little chat. Nobody else needs to know. It feels practically like a private conversation.
You type in a client's name. A few financial details. Maybe a sensitive email you need help wording. Just between you and your AI, right?
Well. Not exactly.
Here's what a lot of business owners don't realise about the tools they're using every day.
ChatGPT
Your conversations are saved by default. Ad personalisation is also switched on by default, using your past conversation themes to target you with ads. You have to actively go into settings to turn it off. OpenAI may not call it selling your data. But businesses are paying to reach you based on what you've told your AI. Think about that for a moment.
Microsoft Copilot
A little better, but conversations are stored for 18 months by default and can be used for AI training unless you opt out. The controls are there, but they are not switched on for you automatically. You have to know to look for them.
Claude by Anthropic
Better privacy commitments at the time of writing. Anthropic does not use your conversations to train their models by default and have committed to remaining ad-free. But it is still worth staying informed, because policies can and do change.
So not all AI is equal when it comes to your data. And that matters, because the information you share is often not just yours.
Think About Who Is Really in That Conversation
When you type a client's name into an AI tool, or paste in a sensitive email, or describe a tricky situation involving someone else, you are sharing their data too. Data they have not consented to share with a third-party platform. Under UK GDPR, that is your responsibility.
And then there are AI apps designed specifically for sensitive sectors. There are therapy and mental health apps now built on AI, where people are sharing deeply personal things: their mental health, their relationships, their trauma. Things they would once only have said in a confidential room to a qualified professional. That data is being stored, potentially used for training, and in some cases shared with third parties. None of the confidentiality protections that come with a human therapist automatically apply. It is a significant concern, and one that the industry is still catching up with.
Be careful what you leave in the hands of AI, it may delete all your emails or your work.
Who Owns What You Create With AI?
That is also worth checking. The ownership rules around AI-generated content are not straightforward, and they vary by platform. If you are producing work for clients using AI tools, it is worth understanding what the terms actually say about who owns the output.
AI Is a Tool, Not a Colleague
People who you would have brought in to help before AI came along, a VA, a copywriter, a bookkeeper, are accountable. They are registered, insured, and bound by confidentiality. You know who to call if something goes wrong. With AI, the rules are still catching up.
AI is useful. I use it myself as part of how I work. But it needs to be checked, verified, and have human input behind it. It does not know your clients. It does not know your business. It does not carry the responsibility you do.
It can also stop you thinking things through properly. It is easy to take the first answer AI gives you and run with it, instead of pausing to ask whether it is actually right for your business or your client. That habit of checking, questioning and using your own judgement does not go away just because AI got there faster. If you're looking for reliable, accountable support for your business, take a look at my services page.
It may seem easier to just use AI. But does it give you the same results as handing your work to someone who already knows you and your business, someone you do not have to retrain every time you open a new chat window?
If you would like to talk through how I work with AI tools responsibly as part of my VA services, drop me an email. Take a look at my services page too.