OpenAI Launches Privacy Tool to Outmaneuver Anthropic

OpenAI is previewing a new service called Private Safety Processing, designed to monitor for AI misuse across multiple user sessions without retaining any customer data. The move directly challenges Anthropic's controversial 30-day data retention policy for its covered models, which has alarmed enterprises handling sensitive information. Unlike Anthropic's approach, which allows limited human review of stored data, OpenAI's automated system uses agents to detect abuse patterns across sessions and only alerts the company via a narrow signal if suspicious activity is found. The rivalry intensifies as both companies pursue IPOs amid fierce competition for enterprise customers.