Anthropic plans to give enterprise customers greater control over data used with its advanced AI models while maintaining a 30-day retention requirement, according to a source familiar with the matter. Customers would be able to retain the data on their own cloud computing infrastructure. The proposed policy changes have been in development for months, with more than 100 customers, including Salesforce, involved in developing the system. Anthropic is expected to roll out a new safety system later this year. The company said in June that it would require 30-day retention of all enterprise customer traffic on its more powerful Fable and Mythos models, as well as future frontier models, to help guard against potential cyberattacks using its technology. Bloomberg News first reported the development on Thursday. Rival OpenAI announced on Wednesday a safety system designed to avoid retaining customer data while still identifying potential misuse.