Anthropic Introduces Enterprise Frontier Safeguards for AI Monitoring

Anthropic has introduced Enterprise Frontier Safeguards (EFS), a system designed to monitor suspicious activity while keeping relevant data inside cloud infrastructure controlled by enterprise customers.

Anthropic says EFS will not retain prompts and outputs under its enterprise protections. The system can correlate behavior across multiple sessions, users and accounts because frontier models can be misused for fraud, cyberattacks and other autonomous threats.

Automated monitoring will send flagged issues to the customer for review, rather than giving Anthropic workers access to potentially sensitive data. Anthropic said it developed EFS with more than 100 customers from financial services, healthcare, manufacturing, telecom, law, retail and the public sector.

The system is scheduled to begin rolling out this fall through a phased rollout and sign-up process. As an interim measure, eligible customers will receive zero data retention on Fable 5 and Fable 5.1 until EFS becomes available to them.

Anthropic previously introduced a 30-day retention policy for some models, but regulated industries and enterprise customers objected to the arrangement. The company said EFS changes where monitored data is stored and allows customers to conduct the review themselves.