Anthropic on 5 November 2026 published a package of Claude Enterprise security and admin controls, according to its enterprise security article and the same-day Help Center release notes. Prompt and pre-tool-call inference hooks are generally available. The security center, centralised management at admin.claude.com, Admin API member sign-out, self-serve IP allowlisting, Desktop shadow-IT blocks and EDR inside the Cowork virtual machine are in beta, Anthropic says.
The companion inference hooks overview says hooks are available to Enterprise plans across Claude, Claude Code, Cowork and other Claude Enterprise products, managed by Owners and Primary Owners.
What Anthropic marks as beta versus generally available
Generally available: inference hooks. Prompt hooks send prompts from supported Claude products to a customer-hosted security tool for an allow-or-deny check. Pre-tool-call hooks check tool calls before they run. Anthropic says the check runs inside Claude’s infrastructure rather than only on the employee device, so one setup can cover chat, Code, Cowork, skills, plugins and connected tools for DLP, transcript archival and custom policy enforcement.
Beta — security center: a page in Organization settings that compares chat, Cowork and Claude Code controls with Anthropic’s recommended configuration, shows current versus recommended values, and can download a report for auditors. Anthropic says the center is guidance, not a certification or compliance guarantee. Organizations with the Compliance API can read the same results through that API. Attestations let admins record reviews and accepted risks in one workspace.
Beta — centralised management: a parent-organization console at admin.claude.com for company-wide SSO, SCIM, domain verification, groups and Compliance API keys across Claude Enterprise and Claude Platform organizations under that parent. Setup requires multi-organization approval and cannot be undone, Anthropic warns.
Beta — access controls: Admin API calls that sign a member out of all browser and device sessions without removing them from the organization; self-serve IP allowlists that admins manage without Anthropic tickets; Claude Desktop settings that can block personal or outside Claude accounts on managed computers to reduce shadow IT.
Beta — EDR in the Cowork / unified Claude VM: admins can run their organization’s endpoint detection sensor inside the local virtual machine on macOS and Windows. Sensors run in detection mode — they report activity and do not block it — and cover the on-device VM, not cloud Cowork sessions.
How this sits beside other Claude enterprise posts
TSN has covered Claude Enterprise unified Cowork, Claude Duet, and Claude Haiku 5.5. Those posts concern product surfaces and model tiers. This security package is an Enterprise admin and DLP path — configuration, hooks and endpoint controls — and stays separate from Anthropic’s consumer Personal Finance beta announced the same week.
What GA hooks and beta admin tools change for security teams
The operational shift Anthropic describes is policy enforcement at inference time plus a single place to see configuration drift. Hooks that are generally available let existing DLP vendors inspect prompts and tool calls before Claude continues; the beta security center and admin.claude.com console are meant to cut the hunt across per-organization settings. Shadow-IT Desktop blocks and Admin API sign-out target account sprawl on managed machines without deleting the user from the tenant.
Admin API sign-out, as documented, ends browser and device sessions but leaves API keys, already-running cloud sessions and scheduled routines alone — so a compromised interactive login can be cleared without wiping every automation token. Self-serve IP allowlisting removes a ticket round-trip to Anthropic for a control many regulated tenants already expect. EDR in the local Cowork VM closes a visibility gap security teams raised about isolated on-device shells, while staying in detection mode so the sensor reports rather than blocks.
Limits still apply. Several headline controls remain beta; EDR is detection-only; the security center is explicitly not a compliance certificate; and centralised management setup is irreversible once approved. Whether large tenants turn hooks on for every product surface — or keep them on Code and Cowork only — will show in later admin documentation and vendor integration notes rather than in this launch article alone.
Sources
- More visibility and control over Claude Enterprise for security teams (Anthropic, 5 November 2026)
- Release notes (Claude Help Center, 5 November 2026 entry)
- Inference hooks overview (Claude Help Center)
Related stories
- Anthropic Makes Unified Claude Chat and Cowork Live for Enterprise
- Anthropic’s Claude Duet Agent Will Not Move Money Even With Approval
- Claude Haiku 5.5: Anthropic’s Small Model Gets Much Cheaper and Much Better at Using a Computer
Follow @tsnmediahub on X for breaking AI news as it happens.
Follow on X
