Claude Code Auto Mode No Longer Requires Opt-In Consent
Claude CodeClaude Code v2.1.152 removes the opt-in consent gate for Auto mode. The AI-classifier permission mode that auto-approves safe tool calls and blocks risky ones now activates directly like any other permission mode, with no confirmation dialog required.
Key Takeaways
- The opt-in consent dialog for Auto mode has been removed, allowing it to activate immediately like any other permission mode.
- Auto mode uses an AI classifier, not blanket permission skipping — risky tool calls are still blocked automatically.
- This applies to Max, Team, Enterprise, and API plans — Pro, Bedrock, Vertex, and Foundry users are unaffected.
- Users with existing Auto mode enabled see no behavioral change — only first-time enablement is now frictionless.
- The
autoMode.hard_denysetting still lets admins or advanced users prevent Auto mode from being selected at all. - This marks the end of Auto mode's research-preview friction, completing its transition to a production-ready, first-class permission tier.
Sources & Mentions
5 external resources covering this update
Inside Claude Code Auto Mode: Anthropic's Autonomous Coding System with Human Approval Gates
InfoQ
Anthropic hands Claude Code more control, but keeps it on a leash
TechCrunch
Auto mode for Claude Code is now available on Enterprise and API
X
Choose a permission mode
Claude Code Docs
Claude Code Updates by Anthropic - May 2026
Releasebot
Auto Mode Now Activates Without an Opt-In Gate
Since its introduction in March 2026, Claude Code's Auto mode has required an explicit opt-in: users cycling through the permission modes with Shift+Tab would hit a consent dialog before Auto mode engaged. That friction is now removed in v2.1.152.
What Auto Mode Does
Auto mode is Claude Code's permission layer that sits between the traditional "ask every time" mode and the risk-heavy --dangerously-skip-permissions flag. Instead of pausing on every file write or bash command, an AI classifier evaluates each tool call in real time:
- Safe operations (reading files, running tests, installing packages) proceed without interruption
- Risky operations (touching unrecognized infrastructure, escalating scope beyond the stated request) are blocked and surface a prompt
The result is fewer interruptions than default mode with meaningfully stronger guardrails than skipping permissions entirely.
The Change: Opt-In Gate Removed
Until v2.1.152, enabling Auto mode for the first time required accepting a consent dialog that explained how the classifier works. This was a deliberate safety measure from the research preview era. With Auto mode now fully rolled out across Max, Team, Enterprise, and API plans, Anthropic has removed this friction point — Auto mode activates immediately when selected, matching the behavior of every other permission mode.
Users who specifically want to block Auto mode from being selected can still do so via the autoMode.hard_deny setting in managed configurations.
Who This Affects
Auto mode is available on Max, Team, Enterprise, and API plans. Pro, Bedrock, Vertex, and Foundry users are not yet eligible. The removal of the consent gate only affects users on eligible plans who had not yet activated Auto mode for the first time — for users who already accepted the previous consent prompt, there is no behavioral change.