Anthropic's Compliance API extends to Claude in Chrome session transcripts

Anthropic has extended its Compliance API so enterprise administrators can retrieve transcripts of employee sessions with Claude in Chrome, the company's browser-based Claude agent, closing a visibility gap for organizations that had only been able to audit chat and API usage.

What's new

According to Anthropic's official Claude Platform release notes, dated September 18, 2026: "The Compliance API local session endpoints now also return transcripts of Claude in Chrome sessions (product_surface value claude_in_chrome), in beta for Claude Enterprise organizations, with your existing Compliance Access Key and the read:compliance_user_data scope."

Concretely, this means an organization that has already set up Anthropic's Compliance API — which lets designated administrators pull records of how employees are using Claude, for audit and data-governance purposes — can now retrieve the same kind of record for sessions run through Claude in Chrome specifically, identified by the new claude_in_chrome surface tag. No new credential type is required: existing Compliance Access Keys with the read:compliance_user_data scope cover the new endpoint. The feature is in beta and limited to Claude Enterprise plan customers.

Context

Claude in Chrome is Anthropic's browser extension that lets Claude take actions inside a user's browser — reading pages, filling forms, navigating sites — on the user's behalf, distinct from the standard Claude.ai chat interface or direct API calls. As Anthropic has pushed Claude further into agentic, action-taking modes across surfaces (the browser extension, Claude Code, the Claude Cowork/Docs/Slides work products, and Managed Agents), its Compliance API has been extended piecemeal to keep pace, adding coverage for each new surface as it matures from experimental to enterprise-supported. This release notes page has logged several such expansions through September, including on-demand conversation compaction for the Messages API and a server-evaluated auto permission policy for Managed Agents, reflecting a steady cadence of incremental platform hardening around Claude's growing set of agentic products.

Why it matters

Browser agents raise a distinct compliance problem that chat interfaces do not: an agent acting inside a browser can read and interact with whatever a logged-in employee can see, from internal dashboards to customer records to competitor sites, and until now that activity sat outside the same audit trail Anthropic already provided for chat and API usage. Extending Compliance API visibility to Claude in Chrome specifically means regulated customers — the same banks, healthcare providers, and public-sector organizations that require Compliance API access before approving any AI tool for employee use — can now bring browser-agent activity under the same data-governance umbrella as everything else Claude does on their behalf. That is a prerequisite, not a nice-to-have, for many enterprises before they will let employees run an agentic browser tool at all, so this beta is likely to unblock Claude in Chrome adoption inside exactly the customer segment Anthropic has been targeting with its Enterprise plan.