Is my business data safe if we start using AI?
It can be, and it mostly comes down to which version of each tool your team uses and how it's set up. The business plans of ChatGPT, Claude and Gemini don't use your business data to train their models by default, while the free and personal plans can unless someone switches that setting off, so the first thing I check is which plans your team is actually on.
Personal plans and business plans
The first thing to check is whether people in your team are using the personal version of a tool for work, because anyone who signs up for ChatGPT, Claude or Gemini with their own email is on a personal plan, and those plans can use their conversations to train the models unless the setting is switched off. The business plans, like ChatGPT Business or Enterprise, Claude Team or Enterprise, and Gemini in Google Workspace, don't train on your business data by default, and they give you an admin view of who's using what.
What I check before we connect anything
- Which plans your team is on, including any personal accounts people have started using for work.
- What each tool can see. I connect tools to your email, files and systems with the narrowest access each one offers, and tell you before we start if a tool would see more than the job needs.
- Where your sensitive information lives. Client records, health information and financial documents may need extra care, or a rule that they stay out of AI tools altogether.
- The settings. Training switched off wherever it's an option, history and retention set to suit you, and sharing turned off for anything that shouldn't leave the business.
Your files stay in your own accounts, and the AI tools we use work with them there. Like any software your business runs on, each AI provider has its own data and privacy policies, and before we connect anything I'll flag any concerns about what a model would see, so you decide what goes in.
If you're in a regulated industry
Health, legal and financial businesses have extra rules about client information, so I set up AI tools with those rules in mind and work with your compliance team or adviser on what's allowed. In Australia, the Office of the Australian Information Commissioner has published guidance on using commercially available AI products, and it recommends, as a matter of best practice, that businesses don't enter personal information, and particularly sensitive information, into publicly available generative AI tools.
Related
- AI consulting and training for leaders and their businessesAI coaching for leaders on second brains, context and strategy, plus automation for quoting, invoicing and reporting that still needs a person to push along.Service
- Should I use ChatGPT or Claude for my business?The better choice depends on your work and the software your team already uses. How I'd compare them on your own tasks, and why many businesses use both.Answer
- Can't I just do this myself with ChatGPT?You can, and plenty of people do. Where doing it yourself works well, where people tend to get stuck, and what working with someone adds.Answer