Require providers that don't train on your prompts
Turn on Require no training in the Data group (data.requireNoTraining) and requests go only to provider endpoints listed upstream as not training on request content. It narrows the pool less than zero data retention; set both to require both.
It applies on chat completions, messages, responses, embeddings, evaluate, suggestions and the mixed-model classifier, at any level, and can't be lifted below. A model with no qualifying provider returns 403 policy_denied saying so and leaves /v1/models. Receipts and the request page show the data rules a request ran under.
Each request also asks the upstream gateway to refuse a provider that may train, as Require ZDR does for retention. Both follow the current upstream listing, checked every hour: a provider it stops listing as not training, or as zero data retention, leaves routing within the hour.
More in Directives: No training.