Anthropic Expands Claude Compliance API With 28 Enterprise Security Integrations
As AI assistants become part of daily work, enterprise security teams are facing a visibility problem. Employees may use them to discuss sensitive business information, upload internal files or create content that is hard to monitor. Anthropic is addressing that issue for Claude with 28 new security and compliance integrations powered by its expanded Claude Compliance API.
The partner list spans major parts of the enterprise security and compliance stack, including data loss prevention, SIEM and security operations, identity, eDiscovery, AI governance and observability. Anthropic named partners including Cloudflare, CrowdStrike, Datadog, Forcepoint, Fortinet, IBM Guardium, Microsoft Purview, Netskope, Okta, Palo Alto Networks, Proofpoint, Relativity, Rubrik, SailPoint, Snyk, Tenable, Varonis, Wiz, Zscaler and others.
The updated Claude Compliance API allows security and compliance tools to retrieve Claude usage data directly, exposing two main types. For Claude Enterprise, the API can make conversation content, uploaded files and project content available to approved security and compliance systems. It also provides activity events across Claude Enterprise and the Claude Platform, including user logins, admin actions and configuration changes. That split helps explain the range of partners involved: data security and eDiscovery tools can inspect the content moving through Claude, while SIEM, identity and posture management platforms can monitor how the system is being accessed and administered.
These integrations hint that Anthropic is pushing Claude closer to a governed enterprise application than a standalone AI tool. Instead of building separate oversight workflows for AI use, Anthropic said the integrations will make it easier for organizations to connect Claude to existing security and compliance platforms, so Claude usage data can flow into the same dashboards and alerting workflows they use for other applications.
The new integrations also follow Anthropic’s recent public beta launch of Claude Security for Claude Enterprise customers. That product uses Claude to scan code for vulnerabilities and suggest fixes, while the Compliance API is focused on governing Claude usage itself. Anthropic is also adding more security controls around how Claude agents interact with enterprise systems. At its Code w/ Claude London event this week, the company announced that Claude Managed Agents can now operate in self-hosted sandboxes and connect to private Model Context Protocol servers.
With self-hosted sandboxes, tool execution can move into an environment configured by the customer, including the organization’s own infrastructure or a managed provider like Cloudflare, Daytona, Modal or Vercel. Anthropic said the agent loop for orchestration, context management and error recovery remains on Anthropic infrastructure, while the customer’s network policies, audit logging and security tooling can apply to the execution environment.
The company also introduced MCP tunnels in a research preview. The feature is designed to let agents reach MCP servers inside a private network without exposing those servers to the public internet. According to Anthropic, a gateway deployed by the customer makes a single outbound connection, avoiding inbound firewall rules and public endpoints.
This focus on security suggests Anthropic is treating governance and observability as the next major requirements for enterprise AI rollout. As AI assistants and agents are integrated closer to internal data, code and other business systems, performance alone will not be enough to accelerate enterprise adoption. The Compliance API makes Claude usage more observable, the partner integrations bring that activity into existing security workflows, and the sandbox and MCP tunnel features give organizations more control over where agentic work runs and what internal services it can reach.
