Fable enterprise user data won't be retained by Anthropic, but some will be analyzed due to 'substantial evidence' of AI misuse

Anthropic, the creator of Fable AI models, has announced changes to its Enterprise Frontier Safeguards (EFS) policy aimed at monitoring AI usage while preserving enterprise customers' privacy. The company will retain relevant activity data within the customer-controlled cloud infrastructure but will refrain from storing prompts and outputs, aligning with enterprise-grade data protection standards.

Anthropic's primary concern is the potential misuse of its frontier models, including Fable 5 and 5.1, for malicious activities like cyberattacks and fraud. To detect such threats, the company requires correlating behavior across multiple sessions, users, and accounts. However, this monitoring process necessitates retaining some data for a period, which initially caused issues for regulated industries and enterprise customers.

In response, Anthropic introduced a 30-day retention policy for certain models. Yet, this solution was not satisfactory for regulated industries and enterprise customers. To address their concerns, Anthropic developed EFS in collaboration with over 100 customers from various sectors such as financial services, healthcare, manufacturing, telecom, law, retail, and the public sector.

EFS aims to shift where monitored data is stored, better matching the customers' demands. Once the system identifies any flagged issues, it sends the information to the customer, keeping Anthropic workers from accessing potentially sensitive data. Until EFS is available to customers through a phased rollout and sign-up process, eligible customers will receive zero data retention on Fable 5 and Fable 5.1.