OpenAI seeks to one-up Anthropic with new customer privacy protections

A competition is developing between OpenAI and Anthropic over who can provide the best privacy protections for enterprise customer data.

The top 3

  1. OpenAI's New Private Safety Processing: OpenAI is introducing 'Private Safety Processing' for its advanced models, which enhances AI security while maintaining Zero Data Retention (ZDR) by analyzing risks through limited signals rather than direct content access, and offers customer-controlled encryption keys for stored data.
  2. Anthropic's Evolving Data Use Policies: Anthropic, initially known for a 'no-training' privacy promise, updated its policy in September 2025 to allow users to opt-in for data use in model training, extending retention to five years if accepted, while a June 2026 update permits sharing conversation data with law enforcement based on 'good faith belief' and may require biometric identity verification.
  3. Apple's On-Device and Private Cloud AI: Apple Intelligence leverages on-device processing to keep user data local for many tasks, while more complex requests utilize Private Cloud Compute (PCC), which processes only relevant data on Apple silicon servers without storing it or making it accessible to Apple, ensuring enhanced privacy.

Sources

Open the full topic