The AI startup will still require enterprise customers that use its most advanced artificial intelligence models to retain data for 30 days, but it will allow them to do so on their own cloud computing infrastructure, according to the report.
Anthropic plans to roll out this new system later this year. It has been working with customers in highly regulated industries for months to develop the option, per the report.
Anthropic did not immediately reply to PYMNTS’ request for comment.
Anthropic announced June 9 that it introduced a new 30-day data retention policy for its Claude Fable 5 and other models with similar or higher levels of capability. The company said that the policy applies to both first- and third-party surfaces and that it will ensure the data’s deletion after 30 days in “almost all cases.”
The company said in its June 9 announcement of the data retention policy: “The data will help us defend against complex and novel attacks (including new jailbreaks and attacks that operate across many requests) as well as help us identify and reduce false positives.”
It was reported on the following day, June 10, that Microsoft was limiting employees’ use of Claude Fable 5 while its legal teams evaluate the changes to Anthropic’s data retention requirements.
The report said Anthropic’s data retention policy includes retaining prompts and outputs for up to two years if they are flagged by Anthropic’s trust and safety classifiers as violating the company’s usage policy.
PYMNTS reported June 11 that a question emerging in C-suite and chief financial officers’ offices was: “What exactly happens to the data flowing through enterprise AI systems, how long is it retained, and what liabilities does that create?”
OpenAI announced Wednesday (Aug. 19) that it is testing Private Safety Processing, a safety system that identifies serious safety risks that may only become visible across multiple interactions, while continuing to offer Zero Data Retention for eligible API customers. The AI startup plans to start rolling out Private Safety Processing in September.