Moves Claude monitoring logs into customers' clouds, unveils enterprise logging option

- Anthropic on September 1 introduced Enterprise Frontier Safeguards, letting Claude monitoring data stay in customer-controlled cloud environments instead of Anthropic’s systems. - Anthropic said the new setup combines zero data retention with automated misuse detection, after business customers objected to retention rules for frontier models. - Enterprise Frontier Safeguards will roll out in phases later this fall across Claude products and supported cloud platforms.

Anthropic on September 1 introduced Enterprise Frontier Safeguards, a new deployment option that lets enterprise customers keep Claude monitoring and activity data in cloud infrastructure they control rather than in Anthropic’s own systems. The company said the system combines zero data retention with automated safeguards designed to detect misuse of its most capable models. Anthropic tied the change to customer feedback as it expanded access to Claude Fable 5.1 and Claude Mythos 5.1, its frontier models. ### What exactly changed in Anthropic’s enterprise setup? Anthropic said Enterprise Frontier Safeguards, or EFS, stores data used for safeguards in infrastructure “controlled by the customer, not Anthropic.” The company said that allows customers to keep logs under their own cloud controls and encryption keys while Anthropic’s systems continue to monitor for misuse. Help Net Security reported on September 2 that the product was built to let Claude logs remain in a customer’s cloud while Anthropic’s automated systems still flag abuse. (anthropic.com) SecurityWeek separately described the system as combining zero data retention with automated monitoring tied to security incidents and misuse detection. ### Why did Anthropic make the change now? Anthropic said it changed course after hearing “a lot of feedback” from business customers about its earlier data retention policy for frontier models. (anthropic.com) CNBC reported the company had introduced that policy in June with Claude Fable 5 and Mythos 5, and that customers in regulated industries objected to how sensitive information would be stored. CX Today reported on September 2 that Anthropic was backing away from the retention requirement for enterprise customers after pushback that the policy made it harder to use Fable 5.1 and Mythos 5.1 with sensitive data. (helpnetsecurity.com) Quartz, via Yahoo, also reported that the new safeguards were designed to replace a policy that had drawn complaints from regulated-industry customers. ### Does this mean Anthropic now offers zero data retention? (cnbc.com) Anthropic said EFS provides “the privacy of zero data retention” while preserving safeguards against adversarial use. In its September 1 product post for Fable 5.1 and Mythos 5.1, the company said eligible customers would be able to use Fable 5.1 with zero data retention until EFS becomes available. The Register reported that zero data retention agreements are not automatic and that customers must apply for them. (cxtoday.com) Unite.ai, citing Anthropic’s announcement, said the offering was designed with feedback from more than 100 enterprise customers and cloud partners, though Anthropic’s own post emphasizes customer, security, compliance and delivery teams rather than naming all participants. (anthropic.com) ### Which products and cloud platforms will get the new option? Anthropic said EFS will be supported on Claude Code, Claude Enterprise, the Claude Platform, Amazon Bedrock, Claude Platform on AWS, Google’s Agent Platform and Microsoft Foundry. The company said the rollout will happen in phases, starting later in the fall of 2026. Anthropic also said the same data-retention change applies to how customers can use its newest frontier models. (theregister.com) In the Fable 5.1 and Mythos 5.1 launch post, the company said EFS would give customers “complete privacy” while keeping safeguards in place, and that the system would be made available to enterprise customers in phases later this fall. ### What does the company say the trade-off is? Anthropic framed the change as an attempt to solve a conflict between privacy controls and frontier-model safety. (anthropic.com) In its announcement, the company said more capable and agentic models raise the risk of misuse and “autonomous misbehavior,” and said it worked with customers to design a system that could preserve both monitoring and stricter data boundaries. Anthropic’s next step is the phased fall rollout of Enterprise Frontier Safeguards across its enterprise products and partner platforms. (anthropic.com) Until then, the company said eligible customers can use Fable 5.1 under zero data retention arrangements, with access subject to Anthropic’s enterprise process. (anthropic.com)

Get your own daily briefing

Scout delivers personalized news, insights, and conversations tailored to your role and industry.

Download on the App Store

Shared from Scout - Be the smartest in the room.