Using AI

Using AI safely: data security and what AI gets wrong

AI is very useful, but it can invent information and leak business data. Seven rules I follow when using AI in export work.

Mr. Dan
Mr. Dan10+ years of export experience

After using AI daily at work for a while, I find it really does save time. But there are risks that export businesses need to understand clearly.

1. AI can "make things up" very confidently

AI sometimes produces information that sounds plausible but is wrong: tariff rates, legal document numbers, company names, market figures. Check official sources for anything you'll base a decision on.

2. AI's knowledge may be out of date

Import rules, freight rates and exchange rates change constantly. Ask the AI how recent its information is, and prefer tools that can search the web and cite sources.

3. Don't put sensitive data into free tools

  • Detailed costs, product formulas, customer lists, full contracts, bank account numbers.
  • Read the data policy of the tool you use. Many services offer business plans that commit not to use your data to train their models.
  • When needed, replace real names with placeholders such as "Customer A" or "Product X".

4. You are responsible for everything you send

Buyers don't care whether AI wrote the email. If it has the wrong price or the wrong commitment, that's your commitment. Always read before sending.

5. Watch out for AI-powered scams

Fraudsters also use AI to write fluent fake emails from customers, and even to fake voices on calls. Any request to change the receiving bank account or make an urgent payment must be verified through a contact channel you already know.

6. Keep your own voice

When every supplier uses AI, identical emails get ignored. Your company's real story and real experience are things AI can't replace.

7. Use AI to learn, not just to do the work

Ask AI to explain why, not just for the answer. Over time, you'll become better at business English and trade operations yourself.