Anthropic Makes Auto Mode the Default in Claude Code for Pro, Max, and Team Plans

Key facts
Date14 Aug 2026

Anthropic has made auto mode the default permission setting in Claude Code for Pro, Max, and Team plans, with the change taking effect for new sessions starting August 14, 2026.

According to the official Anthropic blog post published August 7, users on those plans who have not previously set a different default will begin new sessions in auto mode. Those who already configured a different default may receive a one-time in-product prompt asking whether they want to switch. Any pinned defaults remain unchanged.

How Auto Mode Works

In auto mode, each tool call is routed through a safety classifier before execution. The classifier is designed to block actions that are irreversible, destructive, or directed outside the user’s environment. If a call is blocked, Claude attempts a safer alternative or requests explicit approval. After three consecutive blocks or 20 blocks within a single session, the system falls back to traditional manual approvals.

Permission rules continue to apply before the classifier, with the exception of broad allow rules (such as unrestricted Python execution), which are set aside while auto mode is active. User settings files are left unmodified.

Safety and Performance Data

Anthropic released several quantitative results supporting the change:

  • In a controlled study of 1,053 paid testers, human reviewers caught only 13.6% of deliberately inserted dangerous commands (143 of 1,053). Auto mode blocked 89% (937 of 1,053). Auto mode blocked 800 commands that humans approved; humans blocked only 6 that auto mode allowed. Human detection rates also declined over longer sessions (from ~17% early to ~5% after 50+ prompts), while auto mode performance remained consistent.
  • In real-world production sessions (May–June 2026 data from opted-in accounts), manual approval sessions showed serious unintended harm roughly twice as often as auto mode sessions. At higher severity levels (7+ on a 0–10 scale), the rates were 6.3% for manual versus 2.4% for auto mode.
  • Independent prompt-injection testing by Trajectory Labs (720 attempts across 72 scenarios) recorded a 0% success rate against Claude Fable 5, Opus 5, and Sonnet 5 running in auto mode.

Anthropic also noted that auto mode users on Team and Enterprise accounts have shipped approximately 25% more pull requests in production settings.

Billing and Rollout Scope

Effective immediately with the announcement, Anthropic is no longer charging Pro, Max, and Team users for the additional tokens consumed by the auto mode classifier. The company plans to extend the no-charge policy and the default setting to Claude Enterprise, the Claude API, Claude Platform on AWS, Google Cloud’s Agent Platform, and Microsoft Foundry within the coming month. Enterprise administrators can already set auto mode as the organization-wide default through managed settings.

Anthropic stated that auto mode reduces risk relative to habitual manual approvals but does not eliminate it, and recommended continued human review for high-stakes production changes.

Users can toggle modes with Shift+Tab in the CLI or via the permission dropdown in the desktop application.

Model Releases — the running record
Narracomm · The Record

Model Releases

A dated record of frontier and open-weight model releases — who shipped what, when, under which licence, from which lab and country. The log, not a leaderboard; weighted to open-weight and Chinese releases.

Updated 17 Nov 2026 · Links open our coverage; primaries are cited there
Headline Company DateCoverage
OpenAI Introduces GPT-6.1 Terra with Standard API Prices 50% Lower Than GPT-5.6 Terra and GPT-6.1 Sol OpenAI 28 Oct 2026 Read More →
Alibaba Unveils Qwen Glasses N1 Series at Hangzhou Event with Prices Starting at 2,999 Yuan — 13 Oct 2026 Read More →
OpenAI Launches GPT-6 Sol and GPT-6 Luna to ChatGPT, Treated as High in Cyber and Biology OpenAI 8 Oct 2026 Read More →
DARPA Advances Atom Computing, Diraq, IBM, and IonQ to Quantum Benchmarking Initiative Stage C Microsoft 8 Oct 2026 Read More →
Mishustin: Government will decide when only national foundation models may be used for critical infrastructure; 42 data centres, investment exceeding 280 billion roubles — 8 Oct 2026 Read More →
Reflection Introduces Beam, a 501B Open-Weight Model, Apache 2.0 Weights Later This Month Reflection 8 Oct 2026 Read More →
Mistral opens a public preview of Mistral Large 4, weights due this month — 7 Oct 2026 Read More →
Aleph Alpha Releases Kolibri for Public Administration and Industry Aleph Alpha 7 Oct 2026 Read More →
Council Implementing Decision (EU) 2026/2221 identifies five European Defence Projects of Common Interest under Regulation (EU) 2025/2643 — 1 Oct 2026 Read More →
Huawei rebrands XMAGE as 华为睿影 XMAGE and ties Mate 90 Pro Max 典藏版 and Mate 90 RS 非凡大师 to Kirin 9050 Pro, with a lab claim of 37% and 62% imaging gains Huawei 1 Oct 2026 Read More →
Huawei Lists Mate 90 Series Sale at 12:08 on 1 October 2026, From 5,999 Yuan Huawei 1 Oct 2026 Read More →
Xi ties innovation-driven growth to a sound start for the 15th Five-Year Plan — 1 Oct 2026 Read More →
DeepSeek’s GitHub list puts a plugin harness ahead of its model repos DeepSeek 1 Oct 2026 Read More →
TileLang 0.1.15 adds a Huawei Ascend 950 backend for Python kernel code DeepSeek 1 Oct 2026 Read More →
DeepSeek releases DeepSelect, a TopK kernel for sparse attention and sampling DeepSeek 1 Oct 2026 Read More →
DeepSeek ships Ascend 950 attention kernels and drops Hopper from FlashMLA DeepSeek 1 Oct 2026 Read More →
DeepSeek adds Huawei Ascend backend to TileKernels — 30 Sep 2026 Read More →
CNNIC launches ATI, an AI agent identity layer built on DNS and PKI — 30 Sep 2026 Read More →
DeepSeek Releases DeepEP-Ascend for Huawei Ascend 950 MoE Training — 30 Sep 2026 Read More →
DeepSeek Releases DeepGEMM-Ascend Kernel Library for Huawei Ascend 950 NPUs — 30 Sep 2026 Read More →
Cohere launches Embed 5 embedding models Cohere 30 Sep 2026 Read More →
Musk: SpaceX aiming for Fable- and GPT-6-level model in two to three months — 24 Sep 2026 Read More →
Palantir Adds OpenAI’s GPT-6 Sol and GPT-6 Luna to AIP — 24 Sep 2026 Read More →
Palantir adds xAI’s Grok 4.7 to AIP — 24 Sep 2026 Read More →
Google starts rolling new Connected Apps into Gemini — 23 Sep 2026 Read More →
OpenAI Launches Independent Advisory Group on Mathematics and AI After Internal Model Resolves Navier–Stokes and 100+ Open Problems OpenAI 21 Sep 2026 Read More →
xAI Ships Grok 4.7: Longer-Horizon Coding Model at $2/$6 with New Safeguard Stack OpenAI 21 Sep 2026 Read More →
Lightspeed Launches AI Command Center™ — Conversational AI Built Into Dealer Management System — 18 Aug 2026 Read More →
Stratus Launches Stratus Labs AI Arm and Debuts Agentic Spooling Tools at Innovate 2026 — 18 Aug 2026 Read More →
Anthropic Makes Auto Mode the Default in Claude Code for Pro, Max, and Team Plans — 14 Aug 2026 Read More →
Google Cloud Bolsters Oracle Integration with GA Cloning for Autonomous AI Databases — 16 Jul 2026 Read More →
Windows Insider Announces New Builds for July 6, 2026 Microsoft 7 Jul 2026 Read More →
xAI Makes Grok Build Available in Railway Sandboxes — 2 Jul 2026 Read More →
Microsoft Launches Frontier Company with $2.5B Investment and 6,000 Embedded Experts Microsoft 2 Jul 2026 Read More →
NSA, DEVCOM Army Research Office Launch QuantumEAGLe Initiative — 30 Jun 2026 Read More →
Google Releases Nano Banana 2 Lite to GA and Gemini Omni Flash to Public Preview — 30 Jun 2026 Read More →
Booz Allen Hamilton, OpenAI Announce Partnership for Mission-Ready AI Deployment OpenAI 29 Jun 2026 Read More →
Google Workspace Weekly Recap (June 26, 2026): Incremental Exports, Group Security, and Classroom Read-Along Headline Meaningful Updates Google 27 Jun 2026 Read More →
Microsoft Marketplace Floods with 275 New Offers (June 26, 2026): AI, Compliance, and Industry Solutions Dominate Microsoft 27 Jun 2026 Read More →
BCG: Agentic AI Shifts Transformation from Central Programs to Team-Led Workflow Redesign BCG 19 Jun 2026 Read More →
Google Workspace Adds Gemini Classroom Integration, Longer Veo Videos, Enhanced AI Avatars, and New Admin Chat Controls Gemini 18 Jun 2026 Read More →
McKinsey: AI Ad Economy Shifts: 75% of Advertisers Expect Spend Rise Mckinsey 17 Jun 2026 Read More →
AWS Drops Graviton5 EC2 GA, Gemma 4 Bedrock Availability, FinOps Agent Preview, and CLI v1 Maintenance Mode Signals Amazon 15 Jun 2026 Read More →
Mastercard Launches Agent Pay for Machines — 12 Jun 2026 Read More →
ChatGPT’s Codex Adds Rate Limit Banking and Developer Mode OpenAI 12 Jun 2026 Read More →
xAI Launches Grok Build Plugin Marketplace xAI 11 Jun 2026 Read More →
SKY Co., Ltd. has deployed HPE Private Cloud AI to accelerate secure AI development. — 10 Jun 2026 Read More →
Last updated 14 Aug 2026 · Narracomm · The Record · every figure cites its primary above.
Subscribe
Notify of
0 Comments
Oldest
Newest Most Voted
0
Would love your thoughts, please comment.x
()
x