Best AI for AdvisorsWhat is model training on customer data?
Short answer
When a vendor uses the data you put into its tool to improve or train its AI models, which may expose client information.
Definition
Some AI vendors use customer inputs (prompts, files, recordings, outputs) to train or improve their models. Others promise not to, at least on business plans. The difference is usually in the terms of service, the data processing agreement or an account setting.
Training is different from processing: a vendor has to process your data to give you an answer, but it doesn't have to keep it or learn from it.
Why it matters for advisors
If client information trains a model that other customers use, you've lost control of it. That's hard to square with your privacy obligations and what you tell clients. It's one of the first questions in our vendor due diligence guide.
Get the answer in writing, for the vendor and for any model provider it uses.
Settings can change. Vendors update their terms and add features, and a training opt-out set at signup may not cover a new product. Recheck the setting and the terms at each annual vendor review.
Example
A firm's review finds that a free plan of a transcription app may use recordings to improve its service, while the business plan doesn't. The firm approves only the business plan.
Watch out for
Settings can change. Check training terms at renewal and after major product updates.
Related terms
- Fine-tuningFurther training an existing AI model on a specific set of examples so it performs better at a narrow task.
- Data processing agreementThe contract terms that set how a vendor may use, protect, share, keep and delete the data you give it.
- Sub-processorA third party your vendor uses to handle your data, such as the AI model provider behind a notetaker or CRM assistant.
- Zero data retentionAn arrangement where an AI provider doesn't store your prompts or outputs after returning a response, beyond what it needs to process them.
Learn more
Last updated .