Anthropic has announced a potentially game-changing new feature for Claude Code users on their Claude Team and Enterprise subscription plans: Artifacts.
This update transforms a Claude Code session into a live, interactive, and shareable, custom HTML web page that allows a Claude Code user to enter live code, multiple data sources, and host it in an interactive URL that they can send to other teammates—whether it’s a dashboard, software design, or other product for internal use.
These teammates and the original user can watch the web page update in real-time as Claude Code performs its work autonomously or under user guidance and as related data sources and codebases change.
When Anthropic first introduced Artifacts to its consumer web chatbot in the summer of 2024 (where it evolved from a manual switching feature to a common tool for publishing code snippets and games to the web), it integrated this capability directly into the Claude Code command-line interface (CLI), bridging the gap between engineers and technicians who need a desktop application with deep technology needs. to understand it.
Product and Technology: End of Status Update
At its core, Claude Code Artifacts acts as a dynamic translation layer. Built directly from the continuous context of a user session, the agent uses a local repository code base, associated monitoring tools, and conversational reasoning to spin up specific web pages.
Engineers no longer need to connect to external data sources or connect to temporary infrastructure; AI builds UI from what already exists.
Importantly, these web pages are not static exports. Because the AI works through a terminal session, the open web page refreshes in place, instantly updating graphics and text at the same URL. Each update publishes a new version history, allowing teammates to safely track or roll back an agent’s progress on desktop or mobile.
Battle of Live, Interactive, Shared AI Workspaces: Anthropic’s Claude Code Artifacts Vs. OpenAI Codex Sites
Anthropic’s update has more coming Two weeks after OpenAI released a massive update to its Codex platformoffers a surprisingly similar enterprise hosting feature called "Sites".
This tit-for-tat product rhythm highlights the rapidly growing battle over the enterprise workspace between functions and beyond developers themselves, although there are some important technical and philosophical differences for enterprises to consider.
As revealed on their respective developer documentation web pages, OpenAI builds a platform as a service; anthropic builds a stateless canvas.
OpenAI’s Sites are designed to build sustainable, full-stack web applications. According to the platform’s documentation, Codex Sites host projects that act as Cloudflare Worker-compatible ES modules.
Importantly, Sites supports a persistent backend infrastructure: agents can connect automatically "D1" Relational databases for structured data (such as user progress or saved logs) and "R2" object storage for file uploads. An OpenAI Site can support public logins, integrate with external identity providers, and enable highly specific access controls tailored to specific workspace groups.
It uses a two-step publishing process – saving a reviewable candidate associated with a Git commit before officially deploying it to production. In short, it’s a production environment designed to replace functional in-house SaaS tools.
Anthropic’s Claude Code Artifacts, on the other hand, deliberately avoids the background. The newly released documentation is clear about its limitations: "An artifact is a capture of work, not an application".
Each Artifact is a single, self-contained HTML page enclosed in a specified size of 16 MiB. To ensure organizational security, Claude wraps the published file in a strict Content Security Policy (CSP) that blocks all external network requests. T
means that the page cannot load external scripts, fonts or style sheets and fetchXHR and WebSocket calls are completely blocked. All CSS and JavaScript must be inline, and images must be included as data URIs. Artifacts cannot store form input, call an API during a view, or serve multiple routes.
This technical limitation is actually Anthropic’s deliberate philosophical position: while OpenAI wants to create sustainable software portals for the entire company, Anthropic keeps Claude Code firmly in temporary, highly secure technical workflows. There are Claude Artifacts no software is provided; they are designed to replace whiteboard diagrams, manual error data, and status reports with secure, self-updating visual tools that never leak live data beyond the corporate boundary.
Licensing and Enterprise Security: Keeping the Codebase Private
Because these agents sit at the nexus of proprietary company data and live codebases, licensing and access control are major concerns.
Both Anthropic and OpenAI have opted for closed, proprietary licensing models for these new visual workspaces. For end users and developers, the distinction is very important. Unlike permissive open source software (such as MIT or Apache 2.0) or strict copyleft licenses (such as the GPL)—which give developers the legal freedom to inspect, modify, and self-deploy the underlying code—neither Claude Code Artifacts nor Codex Sites can be independently forked or redeployed.
Enterprise customers do not retain code-level ownership of Anthropic’s rendering engine or Codex’s integration nodes; both operate strictly within themselves managed infrastructures of respective creators.
Both companies have placed a high priority on enterprise security to make this vendor-driven approach affordable for enterprise compliance teams. Anthropic ensures that each artifact is by default private to its author and cannot be exposed to the wider web. When an engineer chooses to share a link, only approved members of their specific organization can see it. System administrators retain ultimate authority by controlling access through organization-level switches, role-based scope, and open retention policies while maintaining control through a centralized compliance API.
OpenAI takes a similar closed approach to Codex Sites, providing the feature mainly for ChatGPT Business and Enterprise workplaces. Like Anthropic, OpenAI relies on system administrators to manage deployment through centralized workspace settings and requires the administrator to explicitly enable Sites through role-based access control (RBAC) for Enterprise levels.
However, because Codex Sites operates more like a hosted web application, its access controls are somewhat more granular. When an engineer is ready to share a hosted URL, they can apply specific access modes: restrict the site to only themselves and workspace admins, open it to all active users in the workspace, or restrict access to individual user groups.
Additionally, to prevent sensitive data leakage, OpenAI provides a custom Sites panel to securely manage runtime environment variables and secrets, ensuring that these keys are not bound to local source files.
Reactions and Reflexes
Applying visual, self-updating UI layers to command-line agents fundamentally changes the way developers view their workflows. As AI handles raw syntax and automates reporting, the stress of communicating technical work to stakeholders is removed.
Boris Cherny, head and creator of Claude Code, emphasized the full benefit of the update. write in X earlier today:
"I use Artifacts in Claude Code for everything: visual explanations of difficult code, system diagrams, quick previews of several animation options, data analysis and dashboards I share with the team," Cherny wrote. "They are a game changer for how I work with Claude. Can’t wait to hear what you think!"
This sentiment is practically on display in Anthropic’s release materials. In one scenario, engineer Claude suggests investigating a user release after a previous software release from Code.
Within seconds, the agent performs a SQL read, builds an interactive drop-down funnel dashboard, and diagnoses whether "Pro accounts stand on the export report". The AI then suggests UI tweaks, updates live graphics as code is reworked, and creates a secure link that a manager can instantly open via mobile.
By turning the terminal into a living, collaborative canvas, Anthropic proves that the most valuable product of an AI coding assistant isn’t just the code itself—it’s context, reasoning, and the ability to instantly share that work.
