Most enterprise Claude deployments stall at the same objection: IT wants visibility. Security wants auditability. Legal wants logs. The answer until recently was “Claude doesn’t plug into your tools that way.” On May 21, that changed.

Anthropic shipped 28 security and compliance integrations alongside a formal Claude Compliance API — a REST API that routes Claude activity into the dashboards enterprise security teams already operate. This is not a dashboard inside claude.ai. This is Claude becoming an event source for your existing SOC stack.

Here’s what that actually means for builders.

What the Compliance API Does

The Claude Compliance API exposes two categories of data programmatically:

Conversation content from Claude Enterprise: the chats, uploaded files, and projects that employees conduct. This is the data your DLP tools need to enforce policy — catching regulated data before it leaves the perimeter, flagging sensitive material being fed to the model, or simply creating a searchable record of AI-assisted work for eDiscovery.

Activity events from both Claude Enterprise and the Claude Platform: user logins, admin actions, and configuration changes. This is the behavioral audit trail your SIEM and identity tools need to correlate AI activity with broader security signals.

Both streams are available via REST, designed to integrate with whatever ingestion pipeline your security team already uses — not a proprietary connector architecture that requires a separate implementation for every vendor.

The 28 Partners, by Category

The integration list spans seven security disciplines:

Data Loss Prevention and SASE: Forcepoint, Netskope, Zscaler, Cloudflare, Microsoft Purview. These tools apply content policy to the conversation data stream — the same tools that scan email, web traffic, and file transfers can now see what’s going in and out of Claude sessions.

SIEM and Security Operations: CrowdStrike, IBM Guardium, ReliaQuest, Sumo Logic, Cribl, Trellix. Activity events flow into the same correlation engines where your team investigates incidents. Claude becomes another telemetry source, not a blind spot.

Data Security: Cyera, Rubrik, Varonis, Tenable, Wiz. These platforms track sensitive data by type and location — they can now classify what’s being shared with Claude and flag policy violations.

Identity Management: Okta, SailPoint. Login and access events flow into your identity platform, enabling you to correlate Claude activity with employee access patterns and apply existing RBAC policies.

eDiscovery and Compliance Records: Relativity, Mimecast, Theta Lake, Proofpoint, Smarsh. For legal, financial services, and regulated industries, Claude conversations become part of the record — searchable, preservable, and producible.

AI Security Posture Management: Geordie AI, Snyk. A newer category that specifically tracks AI tool usage and model interactions as security events, rather than forcing AI activity through tools built for different threat models.

Observability: Datadog, Fortinet. For teams that want Claude usage in their general observability stack alongside application metrics and infrastructure events.

The full partner list: Cloudflare, Cribl, CrowdStrike, Cyera, Datadog, Forcepoint, Fortinet, Geordie AI, IBM Guardium, Microsoft Purview, Mimecast, Netskope, Okta, Palo Alto Networks, Proofpoint, Relativity, ReliaQuest, Rubrik, SailPoint, Smarsh, Snyk, Sumo Logic, Tenable, Theta Lake, Trellix, Varonis, Wiz, Zscaler.

If your enterprise prospect runs any of these — and at most Fortune 500 organizations, several appear simultaneously — Claude now speaks their security language.

The Gap That Matters

The Compliance API has a clearly documented scope boundary: Claude Cowork is not covered.

Claude Cowork is Anthropic’s desktop agent product. Unlike Claude Enterprise’s web interface, Cowork operates with access to the local machine — reading files, executing code, running browser automation. That surface area is significantly larger than a chat session, and it sits entirely outside the Compliance API’s visibility.

Security researchers assessing the May 21 launch called this “half a story.” Their framing is accurate. If your enterprise deploys both Claude Enterprise for general AI work and Cowork for agent-assisted task execution, the Compliance API covers the former and leaves the latter as a blind spot.

For builders, this has two implications depending on where you sit:

If you’re building on top of Claude Enterprise (the API, the managed workspace, not the desktop agent), the Compliance API is immediately useful. Your enterprise sales conversations now have a concrete answer to the governance objection. You can direct IT and security to the partner integrations, point to the REST API for custom ingestion, and treat Claude as a governable data source rather than an unaudited one.

If you’re building something that involves Cowork-style capabilities — or if your enterprise prospect is planning to deploy Cowork — the Compliance API alone does not satisfy an enterprise security review. That gap is currently undocumented in terms of resolution timeline. It’s a question worth asking Anthropic directly, and one worth surfacing with enterprise customers before they discover it during security evaluation.

What the Compliance API Signals

The more significant story is structural, not functional.

Until May 21, the dominant model for enterprise AI compliance was a separate governance layer — tools like Anthropic’s own Model Spec enforcement, third-party AI governance platforms, or internal proxies that intercepted model traffic before it reached employees. These approaches all share a characteristic: they’re adjacent to the AI system, not native to it.

The Compliance API moves Anthropic into a different category. Claude now has native integrations with the same security fabric that governs email, SaaS apps, and developer tools. That’s a different posture than “we have a terms of service and an audit log.” It’s how enterprise software earns a place in security-reviewed environments.

The 28-partner list also suggests Anthropic is treating enterprise security teams as a distribution channel, not just an obstacle. CrowdStrike, Palo Alto, and Microsoft Purview don’t add integrations without customer demand signals. Their presence is evidence that the enterprises Anthropic is targeting are asking for this, not that Anthropic is hoping they’ll ask.

Builder Actions

Three things to do with this information:

1. Map your stack to the partner list. If your enterprise prospect runs CrowdStrike Falcon or Microsoft Purview, Claude now has a named integration with their existing tooling. That removes a barrier that previously required a custom implementation or a policy exception.

2. Get clear on Cowork scope before sales conversations. If your deployment scenario includes any desktop agent capability, do not lead with the Compliance API as a complete governance answer. The gap is real and will surface during security review.

3. Evaluate the Compliance API as infrastructure, not a checkbox. The REST API is programmable — you can ingest Claude activity events into custom pipelines, build internal dashboards, or trigger automated workflows on conversation signals. The 28 named partners are connectors, not the only possible destination.

The enterprise AI governance objection has a concrete answer now. The answer has a documented gap. Know both.

The Claude Compliance API documentation is available through the Anthropic platform developer docs. The 28 partner integrations were announced May 21, 2026.