
Most Claude apps will start new sessions in auto mode next week
On Pro, Max and Team, new sessions will stop asking before each step, so here's what the safety check does and how to keep Manual.
If you're on Claude Pro, Max or Team, Anthropic says new sessions in most Claude apps will start in auto mode next week, so they'll stop asking you to approve each step. Our read: decide now which tasks you want on Manual. Anthropic gives no start date beyond next week.
What changed#
On Thursday at 4:04 PM ET, @claudeai posted: "Starting next week, sessions across Claude web, desktop, and mobile apps will run in auto mode by default on Pro, Max, and Team plans." Anthropic's Oct 22 update to its Claude Code auto mode article words the scope more narrowly: "Auto mode is rolling out next week as the default for new sessions in most Claude apps on Pro, Max, and Team plans." It doesn't say which apps are left out.
| Before | After (per Anthropic) | |
|---|---|---|
| Default in Claude Code (Pro, Max, Team) | Auto since August 14 | Unchanged |
| New sessions in most Claude apps (Pro, Max, Team) | Claude asks you to approve each step (Anthropic's framing) | Auto, rolling out next week |
| Enterprise | Manual by default (help page) | Unchanged for now; admins get an email first |
The Cowork help page, updated Thursday, already reads: "If you have the new Claude experience, the permission setting in the message box offers Auto (default on Pro, Max, and Team plans) and Manual (default on Enterprise plans)." For auto mode in Claude Code, see our Oct 21 story.
Why it matters#
In Auto, per the help page, "Claude reviews each action for safety (such as checking for data exfiltration or prompt injection) and automatically blocks anything it determines to be unsafe." And it reaches a lot: "Auto mode applies to all of your existing connectors, plugins, the built-in browser, Claude in Chrome, and some Cowork actions like fetching websites." Our read: once Auto is on, a connector that can send email or edit a doc can act without a click from you, unless the check stops it.
How to use it#
- In a new session, find the permission setting in the message box. The help page says: "You can change the mode at any time from the mode selector in the chat box."
- Our read: keep Auto for routine, well-defined work, such as tidying files or drafting research.
- Switch to Manual first in the three cases Anthropic's safety page lists, where it calls Manual "Manually approve":
- "The task touches sensitive files, accounts, or sites."
- "You're working with a new tool, plugin, or site for the first time."
- "Mistakes would be hard to undo, like sending messages or making purchases."
- Team owners: the Team and Enterprise page lists two settings in Organization settings > Capabilities that "control "Automatically approve" mode in Cowork". Of Start new tasks in "Automatically approve" mode, it says: "It’s on by default on Pro, Max, and Team plans, and off by default on Enterprise plans. Turn it off to keep "Manually approve" as the default." (Team and Enterprise page) The other, Allow "Automatically approve" mode, removes the mode from members' selectors.
A prompt you can paste at the start of a mixed task (ours, untested):
Before you start: list every action in this task that sends a message, spends money, changes an account setting or deletes anything. Do those last, and stop and ask me before each one.
Our read: it's only an instruction to Claude, not a substitute for Manual.
Good to know#
- The check isn't perfect. The safety page warns: "Either way, if Claude reads malicious content mid-task (a prompt injection), it could act on those instructions before you notice."
- Deletion still asks. "Claude always asks before permanently deleting files, in any mode." (safety page)
- Blocks send it back to Manual. "If Claude keeps running into blocks, it switches back to asking your permission for each step." (help page)
- Auto isn't Skip. Skip all approvals runs with no checks: "Claude doesn't pause to ask and nothing checks its actions automatically." (help page)
- The Team settings are documented for Cowork. The Team and Enterprise page describes them for Cowork tasks only. Our read: Anthropic hasn't said whether they also set the default in the other Claude apps.
- The Claude Code numbers don't carry over. "The data in this article below is specifically for Claude Code." (Anthropic)
- Enterprise waits. "Enterprise admins will get an email before auto mode becomes the default for their organizations." (Anthropic)
What to watch: Our read: when the rollout reaches your account (Anthropic gives no exact day), whether Anthropic names the apps where Auto doesn't become the default, and whether the Team settings cover more than Cowork.
Last checked: Oct 23, 2026, 6:11 to 6:20 AM ET (the @claudeai post, Anthropic's auto mode article and three Help Center pages). What changed: first published Oct 23; no updates since.
Steps
- Open a new session in the Claude app and find the permission setting in the message box.
- Our read: leave it on Auto for routine, well-defined tasks.
- Anthropic's safety page says to switch to Manual when a task touches sensitive files, accounts or sites, uses a new tool, plugin or site, or would be hard to undo.
- Team owners: per Anthropic's Cowork help page, turning off Start new tasks in "Automatically approve" mode (Organization settings > Capabilities) keeps Manual as the default for new Cowork tasks.
Before you start: list every action in this task that sends a message, spends money, changes an account setting or deletes anything. Do those last, and stop and ask me before each one. (ours, untested)