OpenAI’s Private Safety Processing promises to detect abuse across multiple interactions while preserving Zero Data Retention, putting its privacy strategy in direct contrast with Anthropic’s 30-day retention requirement.