If you’re using AI tools regularly for business, it’s worth understanding where the information you put into them actually goes.
Anthropic recently published research on how Australians use Claude. Australia is among the highest per capita adopters of AI tools globally, and what we’re using them for might surprise you.
Coding – the use case most people associate with AI – is well below the global average for Australians. What fills the gap is correspondence, business documents, financial guidance and management tasks. In other words, the information Australians are feeding into AI tools is often sensitive business and client information.
Where your data actually goes
Most people accept AI terms of service the way they accept any software licence – quickly and without reading them. It’s worth knowing the basics.
Free tiers of most AI tools – the no-cost versions of ChatGPT, Gemini and others – typically allow the platform to use your conversations to improve their models. That means content you type into a free AI tool may be reviewed by humans at the company, used in training data, or potentially surfaced in other users’ results in some form.
Paid tiers generally offer stronger data protections, though the specifics vary by platform – make sure you check the specifics on your platform of choice.
The practical implication: if you’re using a free AI tool to draft client correspondence, work through financial decisions or handle any information you’d consider confidential, you may be sharing more than you intend.
Three things worth checking
- Which tier are you on?
Free vs paid makes a meaningful difference to how your data is handled. If you’re using AI tools regularly for business purposes, check what you actually signed up for. - What are you putting in that involves clients?
Client names, contact details, project specifics, financial information – if this is going into an AI tool, check whether your privacy policy covers that use. Most small business privacy policies were written before AI tools existed and don’t address this at all. - What would happen if a client asked?
If a client whose information you’d shared with an AI tool asked whether you’d done that, would you be comfortable answering? That’s a reasonable gut check for whether a particular use is appropriate.
This doesn’t mean stop using the tools
AI tools are genuinely useful and the time savings are real. The point isn’t to avoid them – it’s to use them with a bit of thought about what’s going in and which platform you’re trusting with what information.
For most small businesses, this doesn’t require a policy or a legal review. It requires about ten minutes of reading the terms of service for the tools you’re already using, and a quick check of whether your privacy policy needs updating.
If you’d like help working out which AI tools are worth using for your business and how to set them up properly, the AI Adoption service covers exactly this. For more on your privacy and data obligations as a small business, the Cyber Security & Safety service page is a useful starting point.
