Anthropic Plans to Tweak Data Retention Rules After Enterprise Concerns

Anthropic, the AI startup, is set to alter its data retention policy for enterprise customers, Bloomberg reported on August 20. The company, which requires customers of its most advanced AI models to retain data for 30 days, will now allow them to store this information on their own cloud computing infrastructure. The updated system is scheduled to be implemented later in the year after months of collaboration with customers from highly regulated industries.

In June, Anthropic unveiled a new data retention policy for its Claude Fable 5 and similar models. The policy mandates the deletion of data after 30 days, except in instances where it is crucial for defending against complex attacks and identifying false positives. However, the company has been granted an exception to retain data for up to two years if flagged as violating its usage policy by Anthropic's trust and safety classifiers.

The new policy has raised concerns among enterprise customers, particularly those in highly regulated sectors. Microsoft temporarily halted its employees' usage of Claude Fable 5 while its legal team assessed the changes to the data retention requirements. Anthropic did not immediately respond to inquiries regarding the policy modifications.