OpenAI Previews Private Safety Processing

OpenAI is keeping Zero Data Retention for frontier models while previewing "Private Safety Processing" — a safety layer meant to flag risks across agentic, multi-step tasks without OpenAI staff seeing the underlying content. This post has pulled 2,397 likes and 154 retweets so far.

Open original source ↗Generate a post from this angle

For writing research only. Verify the original source before publishing; market data is not investment advice.