Salesforce's New Consent Tool Is a Bet That Privacy Infrastructure Is the Next AI Bottleneck

Salesforce has launched Global Consent Manager, a centralized, audit-first hub designed to capture and manage consent data at enterprise scale — described as handling billions of consent records on demand without performance bottlenecks. The scenario Salesforce frames the launch around is specific: a customer opts out of third-party data sharing and AI-driven profiling, and traditional consent systems fail to propagate that signal instantly across marketing databases, data lakes, and service dashboards — a gap that becomes a compliance risk the moment an AI workflow touches that data before the opt-out syncs. Core capabilities include centralized consent visibility and enforcement across CRM and Data 360, an immutable audit trail tracking every consent change against global privacy law, and real-time consent-signal sharing enterprise-wide via API and events. Existing Privacy Center customers get access at no additional cost.

The problem Salesforce is solving from the CRM side is the same one OpenAI has been addressing from the model side since mid-August: OpenAI’s Zero Data Retention extension to frontier models, paired with a new Private Safety Processing capability, guarantees eligible API customers’ prompts and responses aren’t retained or reviewed by OpenAI staff, while still allowing automated pattern-detection across related interactions for abuse monitoring — an attempt to reconcile data-retention and security requirements for models handling longer, multi-step agentic tasks. Both moves point at the same emerging category: as agentic AI workflows touch customer data continuously rather than in single transactions, the infrastructure that enforces what a customer has and hasn’t consented to has to move at the same continuous pace, not the batch-cycle pace consent systems were originally built for.

For GTM and RevOps teams, the practical stakes are direct: Salesforce is positioning consent and privacy infrastructure as a prerequisite for scaling agentic AI in customer-facing workflows, not a compliance afterthought layered on once the AI use case is already live.