Customer and company information is pasted into public chatbots every day, usually by someone trying to get their work done faster. Most businesses never decided that should happen.
How it happens
Public AI tools are free or cheap and one click away. Without an approved alternative, people use what is in front of them. A contract goes in to be summarised, a customer list to be sorted, a set of figures to be explained.
The person doing it is rarely careless. They were given a faster way to work and no guidance about where the line is.
Why a ban rarely works
People work around bans. They use a personal account or a phone, and the business can no longer see what is being shared. The practical fix is to give people an option that is just as easy to use and that you control.
Not everything needs the same protection
Treating every piece of information as secret makes AI unusable. Sort it into levels:
- Public. Marketing copy and published information can go anywhere.
- Internal. Day-to-day working material belongs in a business AI service that promises not to learn from your data.
- Sensitive. Customer records, contracts and financials may need AI that keeps data in your own country, or that runs entirely on your own systems.
The trade-offs
Data stored in your country by an overseas-owned provider is not the same as AI you fully control. AI that runs entirely on your own systems keeps everything inside, but it is generally less capable than the leading online services. The right choice depends on what you are protecting and what you can spend.
Steps to take
- Find out which AI tools are in use and what is going into them.
- Sort your information into what can leave and what must stay.
- Provide an approved tool for each level.
- Write a short policy that people will actually read.
How we help
We set up private AI at the level your information needs, and give your people clear rules and a tool that is easy to use. Read more about private AI.