AI Data Training Policy

We do not use Customer Content to train, fine-tune, or improve AI models, ours or anyone else's, except on your documented instructions.

We use AI model providers under business terms that prohibit training on our customers' data, and we enable no-training and restricted-retention options where the provider offers them.

When an AI feature runs, the relevant content is sent to the model provider, a response is generated and returned, and the provider does not retain the content for training. Persistent storage of your data stays within Fulcrum's infrastructure.

Vector embeddings created for search and retrieval are stored in Fulcrum's own database, scoped to your tenant, and are not shared with model providers.

We may use fully anonymized, aggregated metrics that cannot identify any person or business to understand and improve the Service. This never includes readable Customer Content.

Customers can request a summary of our current model providers and their data-handling configurations at [email protected].

Any change that weakens this commitment will be notified to customers at least 60 days in advance, with a clear way to object or export data before it takes effect.