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

OpenAI has unveiled a new customer privacy protection measure aimed at outshining its rival Anthropic. The tech giant is previewing a service called Private Safety Processing, designed to monitor for potential misuse while ensuring no customer data is retained. This new system, which is currently being tested with select customers, is a response to Anthropic's data retention policy, which has drawn criticism from some enterprises handling sensitive information.

OpenAI's privacy-centric approach differs from Anthropic's, which keeps user data for 30 days for safety analysis, a policy that has raised concerns among businesses. OpenAI's existing Zero Data Retention (ZDR) policy already allows customers to scan for abuse without data retention, but Private Safety Processing takes this a step further by analyzing inputs and outputs across multiple conversations.

This long-horizon safety monitoring can detect malicious use of AI that spans several sessions, a capability not possible with the current ZDR system. If triggered, the system sends a narrow signal to OpenAI, which can then decide on necessary enforcement actions, including reaching out to the customer for further clarification.

In contrast, Anthropic's approach involves human review of customer data under strict conditions, with review sessions recorded in a tamper-proof log. The competitive landscape between OpenAI and Anthropic is currently heated, with both companies vying for an edge over each other. Recent reports indicate that OpenAI's Q2 growth has slowed compared to Anthropic, whose annualized revenue run rate is estimated at $65 billion.