Can a competitor end up with your data?

1. 1The dataset is resold. Some standard terms let the buyer sublicense. Without a limit on who, your records could reach a party you would never have chosen.
2. 2The model learns the method. Your data trains or tests models that anyone can use, including competitors. After learning from thousands of your quotes, a model may quote better in your trade, though it never reveals the files themselves.
3. 3Something slips through cleaning. Buyers replace client names, but supplier names, prices and margins may not count as personal data and can survive the scrub.

The promises vary. Some say data is never sold to your competitors. Others restrict it to AI developers listed in your contract, promise never to name you without written permission, or commit to protecting trade secrets. Plenty say nothing in public, so put the question to them.

• Hold back what gives you an edge: pricing spreadsheets, bid comparisons, margin reports and anything a competitor would love to read
• Limit sublicensing to named AI labs in the contract
• Ask for a written no-competitor clause
• Review the cleaned sample yourself for prices, supplier names and margins before release

Excluding folders is normal. See leaving folders out.