Answering a question is different from training a model
An AI provider needs to process a question to generate an answer. That does not, by itself, mean the question is being used to train a model. It also does not mean that no records or technical data are retained.
The Drift beta
The beta uses Cloudflare Workers AI. Cloudflare states that it does not use Workers AI customer content to train its hosted AI models or improve services without explicit consent. See Cloudflare’s data usage documentation.
Drift’s beta saves conversation history and matching details for up to 30 days after the conversation starts, including escalations to the Drift team’s private review inbox. Use fictional scenarios only; this is a test service, not a place to upload personal documents.
How workflows improve
We record limited operational signals such as workflow stage and review reason. Private improvement reports contain counts, not your messages or email address. Session-linked signals are deleted with your conversation and expire after the same 30-day period. Automated regression tests use fictional cases. Raw chats are not automatically used to fine-tune a model or rewrite legal and privacy rules.
Your information
For details about how information from the existing Drift service is processed and retained, read the Privacy Policy or contact Drift before sharing sensitive information.
What can I control?
Share only what is needed, ask how a service will use your information and request access, correction or deletion through the privacy contact route.
