Does AI Train on My Data?

To understand how your data is used, it is important to distinguish between training and inference.

Training

The process of building or improving an AI model using data.

Inference

When a trained model processes your input to generate a result.

In typical AI usage, models process your data to generate responses or results (inference), rather than continuously retraining themselves from every interaction.

Some AI services provide optional capabilities that allow models to improve based on customer data. In these cases:

• Such features depend on the specific service and configuration.
• Customers decide whether these capabilities are enabled where applicable.
• Their use is governed through service-specific settings, agreements, and policies.

This helps ensure that additional data usage is managed through defined controls and configurations.

In generative AI scenarios, SAP may use external foundation models. This raises an important question: how is your data handled in these cases?

SAP enforces contractual and technical controls:

• Customer data is not used for model training or retraining by external providers.
• Data is processed only to fulfill the request.
• Access is governed through SAP-managed services.

Requests can be routed through SAP-managed layers, where controls such as data minimization, masking, and filtering are applied to help protect sensitive information before interacting with external models.

Processing your data does not automatically mean that it is stored long-term or used for training. In many AI scenarios:

• Data is processed during the request to generate a response.
• Retention depends on the specific service and its configuration.
• Customers remain responsible for managing data storage, governance and deletion within their environments.

• SAP Trust Center – Data Privacy: https://www.sap.com/about/trust-center/data-privacy.html
• SAP Trust Center – Security: https://www.sap.com/about/trust-center/security.html

Using AI does not automatically mean that customer data is used to train models. In SAP AI scenarios, most interactions involve inference, where data is processed to generate results.

Some AI services may include optional learning or feedback capabilities, depending on the specific service and configuration. When external AI models are used, customer data is not used by external providers to train their models, and SAP-managed services apply governance and security controls to help protect sensitive information. Understanding the difference between training, inference, processing, and retention helps clarify how customer data is handled in AI environments.