Anthropic adds privacy controls for Claude agents

- Anthropic on May 19 added two privacy and security controls for Claude Managed Agents, expanding the administrative settings available to enterprise customers. - Anthropic’s enterprise materials say customer prompts and responses are not used for model training by default, and retention settings are configurable. - Anthropic’s Help Center says Enterprise customers can choose payment methods at setup, and ACH activation can take up to five business days.

Anthropic on May 19 added two new privacy and security controls for Claude Managed Agents, extending the set of administrative options it offers companies deploying Claude in enterprise settings. 9to5Mac reported the changes as part of Anthropic’s ongoing build-out of Managed Agents, its hosted service for longer-running agent tasks. Anthropic has described Managed Agents as a Claude Platform service that runs long-horizon agents on a customer’s behalf through stable interfaces. Anthropic’s enterprise product materials separately say Claude Enterprise is built around governance, data controls and admin infrastructure for IT and security teams. ### What exactly is Anthropic changing in Claude Managed Agents? 9to5Mac reported that Anthropic introduced two new privacy and security features for Claude Managed Agents on May 19, giving users more control over security and privacy in enterprise deployments. The report said the changes are aimed at enterprise administrators managing Claude agents inside organizations. (9to5mac.com) Anthropic has not, in the sources reviewed, published a full standalone announcement page naming those two controls in detail. But Anthropic’s recent product and help documentation shows the company has been adding more granular enterprise controls around agentic features, including permissions, retention settings and feature access by group. ### How do Managed Agents fit into Anthropic’s broader product lineup? (9to5mac.com) Anthropic said on April 8 that Managed Agents is a hosted service in the Claude Platform for “long-horizon agent work.” In that engineering post, the company said the service separates the session, harness and sandbox so the underlying implementation can change without breaking the external interfaces customers use. Anthropic’s enterprise product page says Claude Enterprise is intended for organization-wide deployment with governance, data controls and admin infrastructure. (support.claude.com) The company says employees can sign in through an existing identity provider, administrators can manage access and offboarding in one place, and retention is configurable. ### What privacy commitments is Anthropic already making to enterprise customers? (anthropic.com) Anthropic says on its Claude Enterprise page that customer prompts and responses are not used to train its models by default. The same page says retention is configurable and that administrators get controls to govern access. Anthropic’s Help Center says Enterprise roles also include permissions around data and security controls. (anthropic.com) In a roles-and-permissions article updated this week, the company says Enterprise owners and primary owners can manage data retention controls, feedback settings, SSO and audit-log requests. ### Are there other admin controls around Claude’s agent-style features? Anthropic’s Help Center says Enterprise organizations can set role-based permissions so specific teams or groups get access to specific Claude capabilities rather than giving all users the same permissions. (anthropic.com) A release-notes entry says admins can organize users into groups, assign custom roles and turn Claude Cowork on for specific teams while restricting features by department. (support.claude.com) Anthropic also says Team or Enterprise plan owners can turn off web search for Cowork and Chat in organization settings. In documentation for Cowork, the company says deleted tasks are removed from task history immediately and deleted from backend storage systems within 30 days, in line with retention periods. ### What does this mean for companies buying Claude Enterprise? (support.claude.com) Anthropic’s Help Center says organizations choose payment methods when creating an Enterprise organization, and ACH activation can take up to five business days after payment is received. That timeline appears in the company’s Enterprise-plan documentation, which 9to5Mac cited alongside its report on the new Managed Agents controls. (support.claude.com) Anthropic’s next published enterprise updates are likely to appear in its Help Center, release notes and enterprise product pages, where it has been documenting new role controls, retention settings and agent-related capabilities. (support.claude.com) (9to5mac.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.