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

OpenAI has announced a new service called Private Safety Processing that monitors customer interactions for potential misuse while retaining none of the customer's data. This approach directly contrasts with a rival company's recently announced policy that retains user data for 30 days to detect safety issues, a practice that has concerned enterprises handling sensitive information. The new system uses automated agents to detect malicious activity across multiple conversations without human review of user data, though it can alert OpenAI to specific types of concerning activity that may trigger enforcement actions. This move reflects intense competition between the two companies, as both seek advantages in the enterprise AI market.