OpenAI Launches Private Safety Processing to Challenge Anthropic's Data Policies

OpenAI has introduced a new service called Private Safety Processing, designed to monitor for AI misuse while retaining no customer data, as TechCrunch first reported. The system expands upon Zero Data Retention (ZDR) by utilizing agents to conduct long-horizon safety monitoring. This allows OpenAI to analyze inputs and outputs across multiple conversations to detect bad actors—such as those attempting to engineer malware—who may spread requests across sessions to avoid detection.

If the system is triggered, it sends a "narrowly defined signal" to OpenAI. The company then determines if enforcement is necessary and may contact the customer for more context; any subsequent data sharing remains at the customer's discretion, an OpenAI spokesperson noted.

This move is a direct response to Anthropic's data-retention policy announced in July. TechCrunch reports that Anthropic retains all sessions and conversations for 30 days for "covered models," which include all Mythos-class models and future models with similar capabilities, such as Fable. While Anthropic states that human review of this data occurs only via a controlled access path with approved reviewers and tamper-proof logs, the policy has concerned enterprises handling sensitive data.

The competition arrives as both firms eye potential IPOs. TechCrunch reports that Anthropic's annualized revenue run rate has reached $65 billion, with investors suggesting a potential $2 trillion IPO valuation, while OpenAI's Q2 growth was reportedly slower than that of Anthropic.