Zero Data Retention Policy

What it is

This provision states that prompts sent to third-party LLMs and the generated responses are never stored by those models and are not used to train them, with the document asserting that data processed under this policy remains the customer's property.

Why it matters (compliance & governance perspective)

This provision establishes a data retention and use restriction applicable to third-party LLM interactions within the Salesforce platform, which is a material term for enterprise customers assessing vendor data handling practices under GDPR, CCPA, and sector-specific data governance requirements. Compliance teams should verify whether this commitment is contractually codified in the applicable Data Processing Addendum, as the document is a policy framework rather than a binding agreement.

Consumer impact (what this means for users)

This provision establishes that customer data transmitted as prompts to third-party LLMs is not retained or used for model training, and the document asserts that such data remains exclusively the customer's property. Enterprise customers relying on this commitment for regulatory compliance purposes should confirm its codification in their specific contractual agreements with Salesforce.

Institutional analysis (regulatory & governance intelligence)

(1) REGULATORY LANDSCAPE: This provision engages GDPR data minimization and purpose limitation principles, CCPA restrictions on secondary use of personal information, and HIPAA requirements for covered entities and business associates regarding use of protected health information. The EU AI Act's provisions on data governance for AI systems may also apply depending on the risk classification of specific use cases. The FTC is the primary US federal enforcement authority for data use commitments made in consumer and enterprise-facing policy documents. (2) GOVERNANCE EXPOSURE: Medium. The provision makes a materially significant commitment regarding third-party LLM data handling, but its enforceability and scope depend on whether it is reflected in binding contractual instruments (MSA, DPA) and whether Salesforce's contractual arrangements with third-party LLM providers impose equivalent obligations on those providers. (3) JURISDICTION FLAGS: EU and EEA deployments create heightened exposure given GDPR requirements for documented data processing limitations and contractual safeguards with sub-processors. California deployments engage CCPA obligations regarding secondary use of personal information. Healthcare and financial services customers face additional sector-specific requirements under HIPAA and GLBA respectively. (4) CONTRACT AND VENDOR IMPLICATIONS: Procurement and legal teams should assess whether the zero data retention commitment is incorporated into Salesforce's standard Data Processing Addendum and whether Salesforce's agreements with third-party LLM providers include equivalent sub-processor restrictions. This provision implies a liability allocation in favor of the customer but does not specify remedies in the event of a breach of this commitment. (5) COMPLIANCE CONSIDERATIONS: Compliance teams should update data processing inventories to reflect this commitment, verify its contractual codification, and assess whether it satisfies applicable data protection impact assessment requirements for AI processing activities. Organizations subject to HIPAA should evaluate whether this commitment, combined with other Trust Layer mechanisms, is sufficient to meet business associate agreement requirements for AI-assisted processing.

Applicable agencies

Provision details

Other risks in this policy

• Model Explainability Provided at Time of Prediction medium
• Salesforce Publishes Model Cards for AI Models medium
• Tools Provided to Identify and Mitigate Bias medium
• Clear Disclosure of AI Terms of Use Provided high
• Customer Data Remains Under Customer Control high
• Customers Must Use AI Responsibly high