Anthropic’s Claude Code Artifacts update brings live, shared dashboards and interactive workspaces to businesses
Anthropic has announced a potentially game-changing new feature for Claude Code users in the Claude Group and enterprise subscription plans: Artifacts.
This update transforms the Claude Code session functionality into a live, interactive, and shareable HTML web page, allowing a Claude Code user to connect live code, multiple data sources, and display it in an interactive URL that they can send to other colleagues – be it a dashboard, application design, or other product intended for internal use.
These team players and the real user can watch the web page update in real time as the Claude Code performs its work automatically or under user guidance, and as the linked data sources and code bases change.
Although Anthropic first introduced Artifacts to its consumer web chatbot in the summer of 2024—where it evolved from a manual conversion feature to a widely available tool for publishing code snippets and games on the web—integrating this capability directly into the Code Claude command-line interface (CLI) and desktop application bridges the gap between the deep engineering needs, who understands
Product and Technology: End of Status Review
At its core, Claude Code Artifacts serves as a dynamic translation layer. Built directly from the seamless context of a user’s session, the agent uses a local code base, connected monitoring tools, and conversational logic to wrap special web pages.
Developers no longer need to connect external data sources or set up temporary infrastructure; AI builds UI on what already exists.
Most importantly, these web pages are not static exports. As the AI works through a terminal session, the opened web page refreshes locally, updating charts and text instantly at the exact same URL. Every update publishes a new version history, allowing colleagues to roll back or track agent progress securely on desktop or mobile.
The Battle for Living, Interactive, Shared AI Work Surfaces: Anthropic’s Claude Code Artifacts vs. OpenAI’s Codex Sites
The Anthropic update comes more than two weeks after OpenAI released a major update to its Codex platform, introducing a surprisingly similar business handling feature called. "Sites".
This tit-for-tat product cadence highlights the rapidly escalating battle in the enterprise workplace across functions and beyond the engineers themselves, although there are important technical and philosophical differences that must be identified by the businesses in question.
As revealed on their developer documentation webpages, OpenAI is building a platform-as-a-service; Anthropic creates a formless canvas.
OpenAI sites are designed to produce long-lasting, full-stack web applications. According to the platform’s documentation, Codex Sites hosts projects that come out as ES modules that are compatible with Cloudflare Worker.
Importantly, Sites support a continuous backend infrastructure: agents can be connected automatically "D1" relational databases of structured data (such as user progress or archived records) and "R2" object storage for file uploads. An OpenAI Site can support social logins, integrate with external identity providers, and allow more specific access controls tailored to specific workplace groups.
It uses a two-stage release process—it saves a review commit linked to a Git commit before it’s officially shipped to production. In short, it’s a production environment designed to replace in-house SaaS tools.
Anthropic’s Claude Code Artifacts, in contrast, deliberately avoids the backend. The newly released documents are silent on its limitations: "An artifact is a capture of a function, not an application".
Each Artifact is a single, self-contained HTML page with a given size of 16 MiB. To ensure the security of the organization, Claude wraps the published file with a Strict Content Security Policy (CSP) that blocks all external network requests. T
his method that the page cannot load external documents, fonts, or style sheets, and fetchXHR, and WebSocket calls are completely blocked. All CSS and JavaScript must be inlined, and images must be embedded as data URLs. Artifacts cannot store form input, call an API at runtime, or provide multiple routes.
This technical limitation is actually a deliberate position of Anthropic’s philosophy: While OpenAI wants to investigate the ongoing software sites of the entire company, Anthropic keeps Claude’s Code focused on highly secure technical workflows. Claude Artifacts not intended to be software; they are designed to prepare whiteboard diagrams, manual bug walkthroughs, and status reports with secure, self-updating visual tools that never leak live data outside of corporate boundaries.
Licensing and Enterprise Security: Keeping the Codebase Confidential
Because these agents reside in the company’s proprietary data connections and live code bases, license and access control is a key issue.
Both Anthropic and OpenAI have chosen closed, proprietary licensing models for these new visual workstations. For end users and developers, the distinction is important. Unlike permissive open source software (like MIT or Apache 2.0) or strict copyleft licenses (like GPL)—which give developers the legal freedom to test, modify, and tinker with the underlying code—Claude Code Artifacts or Codex Sites cannot be independently forked or hosted.
Enterprise customers do not retain code-level ownership over the Anthropic rendering engine or Codex integration nodes; both work strongly within themselves infrastructures managed by the respective creators.
To make this vendor-managed approach attractive to corporate compliance teams, both companies have prioritized organizational security. Anthropic ensures that every artifact is private to its author by default and cannot be made public on the wider Internet. If a developer chooses to share a link, it is only visible to verified members of their particular organization. System administrators retain ultimate authority, manage access through org-level toggles, role-based scope, and clear retention policies, while maintaining oversight through a centralized compliance API.
OpenAI is taking a similar gated approach with Codex Sites, rolling out the feature specifically to the ChatGPT Business and Enterprise workspaces. Like Anthropic, OpenAI relies on system administrators to manage deployments through centralized workplace settings, requiring administrators to explicitly enable sites with role-based access control (RBAC) at Enterprise levels.
However, because Codex Sites works more like a hosted web application, its access controls are very limited. When a developer prepares to share a used URL, they can use specific access methods: restricting the site to themselves and the administrators of the workplace, opening it to all active users in the workplace, or restricting access to custom user groups.
In addition, to prevent serious data leaks, OpenAI provides a dedicated Sites panel to securely manage runtime environment variables and secrets, ensuring that those keys do not need to be committed to local source files.
Reaction and Reflection
The introduction of visual, self-updating UI layers in command-line agents dramatically changes the way developers view their workflow. As AI handles raw syntax and automates reporting, the conflict of communication and technical work for stakeholders disappears.
Boris Cherny, Leader and creator of Claude Code, emphasized the full use of the update in a post on X earlier today:
"I’ve been using Artifacts in Claude Code for everything: visual descriptions of trick code, system diagrams, quick previews of several animation options, data analysis and dashboards that I share with the team," Cherny wrote. "They are a change in the way I work with Claude. I can’t wait to hear what you think!"
This sentiment is literally reflected in Anthropic’s launch materials. In one case, a developer informs Claude Code to investigate user downgrades from previous software releases.
In a matter of seconds, the agent performs a SQL read, creates an interactive funnel dashboard, and diagnoses it. "Professional accounts stand on the export sheet". The AI then suggests UI adjustments, updates live charts as the code is refactored, and generates a secure link that the manager can quickly open via mobile.
By turning the terminal into a live, collaborative canvas, Anthropic proves that the most important output of an AI coding assistant isn’t just the code itself—it’s the context, the thinking, and the ability to quickly share that work.
