Anthropic is reportedly working on a revision of its data-retention policy for business customers, as major AI firms try to balance security controls with enterprise requirements.
Anthropic in June said it would retain all customer data for 30 days for users of its most powerful Mythos and Fable models, and for all future frontier models.
The policy shift was designed to help the company identify and prevent new kinds of cyberattacks using its technology, amid well-publicised incidents in which powerful models unpredictably hacked outside companies without being asked to do so.
Image credit: Dario Lo Presti/Shutterstock
Enterprise data
