What happened
On Sept 29-30, 2026 Anthropic released the Claude Compliance API (endpoints under /v1/compliance/*) letting Claude Enterprise and Claude Platform customers export chat conversations, uploaded files, projects, and Claude Code/Cowork session transcripts (prompts, responses, tool calls) plus an activity feed to DLP, SIEM, SASE, eDiscovery and AI-SPM platforms. First-wave ecosystem integrations include CrowdStrike, Splunk, Microsoft Purview, Proofpoint, TrendAI Vision One, Nightfall, MIND and Pi Security.
Why it matters
This converts otherwise-invisible AI-agent activity (coding sessions, tool calls, connectors, MCP usage) into standard security telemetry with verified user attribution, closing a major governance gap for enterprise AI — the same visibility gap every agent-security vendor has been selling against. It is the first large-model vendor to ship a comprehensive programmatic compliance/telemetry surface for its coding and knowledge-work agents.
Applicability
CISO/security-platform teams running Claude Enterprise or Claude Code should enable the API (Primary Owner role) and wire it into SIEM/DLP posture to audit agent actions; AI-security vendors should build on it. Assess immediately as it is generally available now.