Oracle Analytics and AI

Rebecca Bortone

Hello,

I would like to understand whether Oracle provides any official documentation regarding the data and information used to train Oracle AI capabilities, especially within Oracle HCM Cloud.

In particular, I would like to know:

• which types of data are used to train the Oracle AI models;
• if there is any official Oracle documentation explaining the AI training process, data sources, privacy model, and data isolation mechanisms.

Could you please share any available documentation, white papers, MOS notes, or official Oracle references on this topic?

Thank you in advance.

Rebecca

Arindam Sadhukhan-Oracle

Hi Rebecca,

Oracle has documentation that covers this at the Oracle Fusion Applications AI platform level and also for OCI Generative AI data handling.

For Oracle AI embedded in Fusion Applications, Oracle documents that each organization has a securely partitioned platform instance, customer data and trained predictive ML models are isolated, and there is no sharing or mixing of data between organizations. The same page also states that information is not persisted or used to train models or generate responses for other customers.

A few useful starting points are:

• Oracle AI for Fusion Applications overview: https://docs.oracle.com/en/cloud/saas/fusion-ai/
• Oracle AI Platform for Fusion Applications:https://docs.oracle.com/en/cloud/saas/fusion-ai/aiaqa/what-is-the-oracle-fusion-ai-platform.html
• User behavior, input, and feedback in Fusion AI models:

https://docs.oracle.com/en/cloud/saas/fusion-ai/aiaqa/how-is-user-behavior-input-and-feedback-incorporated-into-the-model.html

• AI governance and standards for Fusion Applications:

https://docs.oracle.com/en/cloud/saas/fusion-ai/aiaqa/what-ai-governance-and-standards-accreditations-does-oracle-have-for-fusion-applications.html

• OCI Generative AI data handling: https://docs.oracle.com/en-us/iaas/Content/generative-ai/data-handling.htm
• OCI Generative AI Agents data handling:https://docs.oracle.com/en-us/iaas/Content/generative-ai-agents/data-handling.htm

At a high level, the answer depends on the AI type:

For predictive/classic AI in Fusion Applications, models may use customer-specific application data and user behavior signals to improve recommendations or predictions, but Oracle documents that an organization’s feedback data is isolated and not incorporated into shared models.

For generative AI/LLM use cases, Oracle documents that the LLM used by Fusion Applications is a foundational model and is not further trained or fine-tuned using customer prompts. OCI Generative AI documentation also states that prompts/inputs are not retained inside OCI Generative AI, and fine-tuning data is used only for the customer’s custom model, not to improve general OCI Generative AI use cases.

For formal security/procurement review, I would also suggest checking Oracle’s AI CAIQ / CSA STAR and Oracle Cloud Compliance materials, since those are intended for governance, risk, and vendor-assessment questions.

Regards,

Arindam.