/
Navigation
Chronicles
Browse all articles
Explore
Semantic exploration
Research
Entity momentum
Nexus
Correlations & relationships
Story Arc
Topic evolution
Drift Map
Semantic trajectory animation
Posts
Analysis & commentary
Pulse API
Tech news intelligence API
Browse
Entities
Companies, people, products, technologies
Domains
Browse by publication source
Handles
Browse by social media handle
Detection
Concept Search
Semantic similarity search
High Impact Stories
Top coverage by position
Sentiment Analysis
Positive/negative coverage
Anomaly Detection
Unusual coverage patterns
Analysis
Rivalry Report
Compare two entities head-to-head
Semantic Pivots
Narrative discontinuities
Crisis Response
Event recovery patterns
Connected
Search: /
Command: ⌘K
Embeddings: large
TEXXR

Chronicles

The story behind the story

days · browse · Enter similar · o open

Anthropic says auto mode will be the default in Claude Code for Pro, Max, and Team plans from August 14, claiming it catches harmful actions better than humans

Auto mode is now the default in Claude Code for Pro, Max, and Team plans (via) Anthropic are really confident in Claude Code's auto mode …

Simon Willison's Weblog Simon Willison

Context & Ripple Effects

Anthropic introduced auto mode in March to let Claude Code make permission-level decisions while blocking destructive actions such as mass file deletion. The new default turns that initial permission-decision feature from an optional capability into the standard experience for paid Claude Code users. It also follows Anthropic's broader effort to have Claude apply principles rather than fixed rules through its reworked constitution.

First-order effects

  • Pro, Max, and Team users will encounter Claude Code with automated permission decisions enabled by default from August 14, placing Anthropic's harmful-action detection directly in their development workflows.
  • Anthropic is taking responsibility for making the automated guardrail the baseline, backed by its claim that auto mode catches harmful actions better than human reviewers.

Second-order effects

  • Teams using Claude Code will need to treat the model's permission judgments as an operational control rather than an elective productivity setting, changing where human review is applied.
  • Anthropic's computer-use and coding products gain a more consistent safety model as the same company moves from exposing agent capabilities to defaulting their automated oversight.

Third-order effects

  • If Anthropic's performance claim holds in routine use, AI coding tools may shift toward oversight inversion: models perform more of the immediate gatekeeping while people audit exceptions and policy outcomes.
  • That shift makes operational assurance a product differentiator, because providers will need to demonstrate that automated agents can act quickly without weakening destructive-action controls.

The trend: AI developer tools are moving from human-approved actions toward model-mediated permissioning, with safety performance becoming the basis for making autonomy the default.

Discussion

  • @trq212 @trq212 on x
    we should have called this post “defeating the lethal trifecta” https://claude.com/... [image]
  • @bcherny Boris Cherny on x
    The team and I use Auto mode exclusively, and have been for many months. I couldn't imagine going back to permission prompts! Really excited to get this out to everyone.
  • @claudedevs @claudedevs on x
    We're making this change for two reasons: 1. In our testing, auto mode matched or beat manual permission review on every safety measure we tracked. 2. It makes long-horizon work more viable. Claude runs longer between interruptions, so you can run multi-hour tasks in the
  • @claudedevs @claudedevs on x
    Starting August 14, auto mode will be the default permission mode in Claude Code for Pro, Max, and Team users. Auto mode reviews shell commands and actions with a separate classifier. In testing, it caught 89% of dangerous commands. Manual approval caught 14%.
  • @namiixbt Nami on x
    Honestly, I think this is a good move The interesting part isn't really the 89% vs 14% number. It's that permission prompts are starting to become a bottleneck for agents that are supposed to work for hours on their own If Claude has to stop and ask me for permission every [image…
  • @the_alex Alex on x
    OpenAI: “We're gearing up to launch our craziest model yet!” Anthropic: “We're adding more guardrails lol”
  • @kunchenguid Kun Chen on x
    “it caught 89% of dangerous commands” well... what about the other 11%?  being better than humans (who usually won't look carefully) doesn't mean it's useful this proves reviewing each command is simply not the right solution instead, we need to secure the environment and set bou…
  • @trq212 @trq212 on x
    automode is much safer than any other permission system out there, especially reviewing them yourself excited to announce we're rolling it out to everyone by default, with no overhead cost for the classifier [image]
  • @kienbuilds Kien Pham on x
    @ClaudeDevs can't wait for my files to be accidentally* deleted [image]
  • @claudedevs @claudedevs on x
    One reason we trust it more than manual approval: in a study with 1,053 paid testers, we swapped a permission prompt for a clearly dangerous command (text only, nothing actually ran).  Testers caught it 13.6% of the time, and closer to 5% after 50 prompts.  Auto mode blocked the …
  • @claudedevs @claudedevs on x
    You'll get an in-app notification when the default flips. If you've already set a default permission mode, Claude will ask before changing anything. You can still switch modes with Shift+Tab anytime. Admins can pin defaultMode (or disable auto mode entirely) in managed settings
  • @st3v3li Steve Li on x
    @ClaudeDevs Do we get a usage reset for this? [image]
  • @sipirtu.com @sipirtu.com on bluesky
    Anthropic will make Auto Mode the default in Claude Code for Pro, Max, and Team plans starting August 14, citing a classifier that catches 89% of dangerous commands versus 13.6% for humans.  Developers may spend less time writing code and more time monitoring AI output.  —  Sourc…
  • @kimiatehrani Kimia on x
    The auto mode update is the kind of change you appreciate more the longer you've used Claude Code Long sessions used to mean staying nearby to approve things as they came up. Now the classifier handles that review layer and the numbers behind it are solid, 89% of risky commands
  • @udiwertheimer Udi Wertheimer on x
    this is great news. i've been using auto mode for months and it's clearly the best permission mode if you ask people to approve everything manually they'll just click approve on everything, having no idea what it means or they'll switch to “approve all dangerously” which is
  • @alliekmiller Allie K. Miller on x
    I use auto mode for everything and now that will be the default in Claude. Anthropic had to decide whether to prioritize maximization of human control or minimization of risk, and it chose the latter. I still recommend to newer users to WATCH Claude Code go through the