Anthropic launches Claude for Google Workspace beta with in-file editing across Docs, Sheets and Slides

Anthropic launches Claude for Google Workspace beta with in-file editing across Docs, Sheets and Slides

N
News Editor
2026-10-07 08:09:35
Anthropic has rolled out a public beta for Claude for Google Workspace, putting Claude directly inside Google Docs, Sheets, and Slides through a sidebar that can read the file a user has open and edit it in place. The feature is currently limited to paid plans, including Pro, Max, Team, and Enterprise. According to Anthropic, users can approve edits one by one under the default review mode or switch to a broader acceptance setting for routine content changes. As of Oct. 7, the extension had been installed by more than 8,000 users on the Google Workspace Marketplace. The launch gives Claude a direct position inside Google’s productivity suite, though the integration still comes with clear limits around permissions, browser support, task duration, comments access, chart behavior, and sensitive actions such as sharing files or changing access rights.

Anthropic on Oct. 7 introduced a public beta for Claude for Google Workspace through Claude’s official X account, adding the assistant directly into the sidebar of Google Docs, Sheets, and Slides and allowing it to edit files in place. The feature is available to paid users on the Pro, Max, Team, and Enterprise plans.

As of Oct. 7, the Google Workspace Marketplace listing showed that the extension had been installed by more than 8,000 users.

Two ways to access Claude inside Workspace

The first route is through the add-on itself. After installing it from the Marketplace, users can open a file, choose Claude from the Extensions menu, and click Open Claude. That places a chat sidebar next to the document. In its official announcement, Anthropic said Claude can read the file currently open, detect selected text, cells, or slides, and make changes directly inside the file.

In Docs, Claude can revise a sentence or change a heading style without disturbing the surrounding layout. Larger rewrites appear as suggestion cards in the sidebar, and the user can apply or skip them.

In Sheets, it can write formulas, create pivot tables, and build native charts. For data-cleaning tasks, Claude can bring Python into the workflow and write the processed output back into the spreadsheet.

In Slides, it creates new pages based on the presentation’s existing theme and then flags elements that overlap or extend beyond the slide boundary.

The default setting is Ask before edits. Under that mode, each change first appears as a summary card, and Claude waits for the user’s approval before making the edit. If the user switches to Accept all edits, routine content changes no longer require item-by-item approval. Anthropic said the edits are recorded under the user’s own name, so collaborators can see the changes, but not the conversation.

The second route works in reverse. A user can paste a Google file link into a Claude chat or ask Claude to create a new Google Doc, Sheet, or Slide. The file then opens in a pane next to the conversation, where the user and Claude can edit at the same time. That pane is currently limited to the web version on Chrome and the desktop app with the built-in browser enabled.

What the integration can and cannot access

Anthropic said the most basic limitation of the add-on is that it can only read the single file that is currently open. Claude’s help center states that, aside from connector content opened by the user, it cannot see other files in Drive, emails, or calendar items. The help center also says some of these limits were left in place deliberately, and lifting them would require broader authorization from Google.

If the sidebar is closed, any task in progress stops. A single operation can run for up to six minutes, which Anthropic says is a Google-imposed limit. Firefox is not supported.

Inside the Docs add-on, Claude cannot read comments. But when a file is accessed from Claude’s side through a connector, comments are available. In Slides, charts generated by Claude are inserted as images and do not update automatically with changes in the underlying numbers.

Prompt injection and restricted actions

Anthropic also pointed to prompt injection as a risk. In that type of attack, instructions aimed at the AI are hidden inside a document, such as white text or extremely small text, to try to make the model follow them. The help center says Claude treats document content as data rather than commands and will flag hidden text.

Even with Accept all edits enabled, actions such as adding links, inserting external images, or removing another person’s access still require approval each time, and there is no always allow option. Claude also cannot share files, grant access to others, or change ownership.

Pressure builds in workplace AI tools

The update leaves Google’s editor and cloud file infrastructure in Google’s hands, but the AI making changes inside those files no longer has to be Gemini. With this beta, Anthropic is pushing Claude into one of Google Workspace’s core use cases, intensifying competition in productivity software.

This article was originally published by Bit.Fan. For more cryptocurrency news and market insights, visit www.bit.fan.
100

Disclaimer:

The market information, project data, and third-party content displayed on this platform are for industry information sharing only and do not constitute any form of investment advice or return commitment.

Cryptocurrency trading carries high risks. Users should fully assess their risk tolerance and make independent decisions. All profits, losses, and legal responsibilities are borne by the users themselves.