# Anthropic added hosted Claude-powered apps to Artifacts Anthropic's July 25, 2025 release let Free, Pro and Max users build, host and share Claude-powered artifact apps, with recipients using their own Claude accounts. Canonical: https://rangoon.ai/insights/archive-anthropic-claude-powered-artifacts-july-2025/ Author: Rangoon Editorial (https://rangoon.ai/insights/editorial/) Published: 2026-10-24 Updated: 2026-10-24 Event date: 2025-07-25 Tags: [Anthropic](https://rangoon.ai/insights/tags/anthropic/), [Claude](https://rangoon.ai/insights/tags/claude/), [AI apps](https://rangoon.ai/insights/tags/ai-apps/) Source note: Historical reporting: Source 0 is Anthropic's dated July 25, 2025 Claude-powered Artifacts announcement. Source 1 is a current, product-specific help record used only for present capability, connector and storage context; it is explicitly not used to claim that current behavior applied at the July 2025 launch. Image: July beta: description to shared Claude app. Original Rangoon explanatory diagram: Source boundary: dated primary release reporting; original editorial analysis. No vendor interface, photograph, measurement result, or endorsement implied. ## Key takeaways - Anthropic's dated post made hosted, shareable Claude-powered Artifact apps a beta capability for Free, Pro and Max users. - A recipient authenticated with a Claude account and used their own subscription allowance, avoiding a creator-supplied API key and usage bill in the announced model. - Current support detail now includes connectors and persistent storage for eligible plans; it is not launch-time evidence. ## A July beta made Artifacts into hosted apps [source 1](https://www.anthropic.com/news/claude-powered-artifacts) On July 25, 2025, Anthropic announced the ability to build, host and share interactive AI-powered apps directly in the Claude app. It said Claude could create Artifacts that interact with Claude through an API, turning those artifacts into apps. The page put the capability in beta for Free, Pro and Max users. It also said a person could describe an app, have Claude write code, iterate through feedback, and share the completed app through a link without a separate deployment process. That is the bounded historical event. The post did not claim that each generated app was production-ready, that a creator had tested it, or that every later Claude plan and artifact format had this exact launch state. ## The shared-app model put identity on the recipient [source 1](https://www.anthropic.com/news/claude-powered-artifacts) Anthropic made a specific operating claim about shared Claude-powered apps. A person using one would authenticate with an existing Claude account, and their usage would count against that person's subscription rather than the creator's. The company said no API keys were required and the creator would not pay for another person's usage. Those mechanics distinguish the launch from handing a shared secret to an application or paying a central API bill for every visitor. They do not establish that a recipient understood an app's code, that an account had sufficient usage allowance, or that the app was appropriate for sensitive input. Authentication answers who consumes a platform entitlement; it does not independently validate an app's data handling, output, or business purpose. ## The announcement named a narrow beta capability [source 1](https://www.anthropic.com/news/claude-powered-artifacts) The July source listed what the beta could do: use a Claude API within an Artifact, process files, create rich React interfaces, and let users see, fork and customize an Artifact. It also listed launch-time limits: no external API calls yet, no persistent storage, and a text-based completion API. These are historical statements about the beta described by the dated post, not a durable definition of every Artifact. The page's examples included games with adaptable NPCs, learning tools, CSV analysis, writing assistants and workflows that orchestrate multiple Claude calls. Examples show intended uses; they do not prove that a particular artifact completed those tasks reliably, handled a user's data appropriately, or had permission to take an action. ## Current help context differs from launch limits [source 2](https://support.anthropic.com/en/articles/9487310-what-are-artifacts-and-how-do-i-use-them) Anthropic's current help article says Artifacts span documents, decks, dashboards and small interactive tools, and says current eligible Pro, Max, Team and Enterprise users can connect apps to an Artifact and store data in one. It also describes plan-specific availability, administrator controls and the requirement to enable cloud code execution and file creation. Those are present-tense product details read on October 24, 2026. They contradict the July post's beta storage and external-call limitations because the product changed. For that reason, they supply current context only and do not revise the date, availability, limits or technical surface established by the July 2025 announcement. ## Generated app code still required an owner and review [source 1](https://www.anthropic.com/news/claude-powered-artifacts) [source 2](https://support.anthropic.com/en/articles/9487310-what-are-artifacts-and-how-do-i-use-them) The July launch lowered the path from a plain-language request to a hosted, shareable Claude-backed app. It did not remove engineering work. A creator still needed to inspect generated code, test interaction and failure paths, review text and asset rights, decide what data an app should receive, and verify the sharing audience. The dated page does not provide a security assessment, a reliability benchmark, a privacy approval or release authority for any specific Artifact. In particular, a recipient's Claude login is not consent to every downstream action an app might appear to suggest. Historical reporting should preserve the product change while keeping output review, data handling and publication governance in their own evidence lanes. ## Sources 1. [Anthropic: Build and share AI-powered apps with Claude](https://www.anthropic.com/news/claude-powered-artifacts) 2. [Anthropic Help: What are artifacts and how do I use them?](https://support.anthropic.com/en/articles/9487310-what-are-artifacts-and-how-do-i-use-them) ## Related reading - [Anthropic launched a Claude directory for one-click connector discovery](https://rangoon.ai/insights/archive-anthropic-claude-connectors-directory-july-2025/) - [Anthropic introduced Claude Research with web and Workspace search](https://rangoon.ai/insights/archive-anthropic-claude-research-april-2025/) - [Anthropic donated Agent Skills to AAIF: treat open governance as a new supply-chain input](https://rangoon.ai/insights/anthropic-agent-skills-aaif-donation/) ## More information - [Documentation index](https://rangoon.ai/llms.txt) - [AI access and policies](https://rangoon.ai/ai/) - [Source repository](https://github.com/hypler-dev/rangoon)