What data you shouldn't share with an AI

"Company leaks customer data to an artificial intelligence tool" is the kind of headline that worries more and more leadership teams. While everyone talks about productivity and automation, one question keeps getting skipped: what information is actually being shared with these tools.

What data protection authorities recommend

Data protection authorities recommend avoiding entering personal data, sensitive information, confidential business data, and third-party data into AI tools without a prior assessment. It is not a blanket ban on using AI — it is a call to think before pasting text into a chat.

Three categories of data to protect

• Personal data of customers and employees — Names, phone numbers, addresses, ID numbers, or emails require special care. It is common for someone to copy a conversation with a customer or upload a spreadsheet with contact data without a second thought.
• Contracts, reports, and internal documents — AI can summarize a contract or extract conclusions from a report, but that does not mean confidential information — contracts, reports, strategies — should be uploaded to a public tool.
• Financial information, strategy, and credentials — Income statements, margins, business forecasts, or launch strategies need protection. And this one is non-negotiable: passwords, API keys, tokens, or credentials should never be entered into an AI chat.

The rule to apply before sharing anything

Before pasting any data into a public AI, apply this simple rule: if you wouldn't send it to a third party without knowing how they will handle it, don't paste it into an AI chat whose terms of use you probably haven't read in full.

For truly sensitive data, the alternative is not to stop using AI, but to use the right AI: local models or private platforms — like Minte — where information never leaves your organization.