OpenAI is launching a new initiative, Private Safety Processing, to monitor for AI misuse across multiple customer interactions without retaining any customer data. This move is a strategic play against competitor Anthropic, whose recent policy of retaining customer data for 30 days for safety purposes has caused concern among businesses, especially those in regulated industries like healthcare and finance.
OpenAI's existing Zero Data Retention (ZDR) policy already prevents the company from retaining prompts or model responses for eligible API customers. The new Private Safety Processing expands on ZDR by enabling cross-session monitoring, allowing OpenAI to detect complex misuse patterns that might not be evident in single interactions. When potential risks are identified, the system sends a high-level signal to OpenAI, without revealing the underlying customer content, and customers can choose to share more information if needed.
This development comes amidst intense competition between OpenAI and Anthropic, with reports indicating OpenAI's Q2 revenue growth lagged behind Anthropic's, whose annualized revenue run rate is reportedly $65 billion. Tech leaders, including White House AI adviser David Sacks and Microsoft CEO Satya Nadella, have criticized Anthropic's data retention policy, emphasizing the need for AI companies to provide certainty that customer data will not be kept.
