DOOMBENCH.COM
Menu
Autonomy and agency

Anthropic gives paid Claude users autonomous browser actions

Anthropic made Claude in Chrome generally available on every paid plan and enabled automatic approval for actions its safety classifier judges consistent with the user's request. Claude can use existing logins to read, type, click, navigate, and fill forms. In a stronger prompt-injection evaluation, probes plus the classifier reduced successful attacks to zero for Sonnet 5, Opus 5, and Mythos 5 and 0.3 percent for Fable 5, with the remaining successes manually rated low severity.

0 comments · 0 votesOpen discussion

Public discussion is readable by everyone. Sign in to comment, reply, or vote.

No comments yet. Start the discussion.

Send detailed or private feedback →
CURRENT ASSESSMENT · REVISION 1
TOWARD DOOM55confidence 92/100

Why it moved the index

This is a broad consumer deployment of browser autonomy through users' existing authenticated sessions, with per-action human approval no longer the default. The measured probes and action classifier materially reduce current prompt-injection success and temper the risk, but Anthropic says the attack surface keeps evolving and Fable 5 was not at zero. Magnitude reflects meaningful diffusion and authority rather than evidence of a real compromise; confidence is high because the dated primary source specifies availability, actions, models, evaluation design, and measured outcomes.

AUDIT TRAIL

Assessment history

  1. R1
    Toward 55 · confidence 92

    Adds a distinct consumer browser-autonomy rollout and new model-specific prompt-injection measurements not present in the durable developer-platform record.

    27 Aug 2026
SHARE THE FINDINGS

Share this page

DoomBench social sharing card for Anthropic gives paid Claude users autonomous browser actions.

  1. DoomBench assesses “Anthropic gives paid Claude users autonomous browser actions” as evidence moving toward doom, with magnitude 55 and confidence 92 out of 100 in the autonomy and agency category.

  2. The DoomBench assessment of “Anthropic gives paid Claude users autonomous browser actions” is based on reporting from Claude by Anthropic and records the editorial rationale, source quality, attribution, and revision history.

  3. DoomBench summarizes “Anthropic gives paid Claude users autonomous browser actions” as follows: Anthropic made Claude in Chrome generally available on every paid plan and enabled automatic approval for actions its safety classifier...

    https://www.doombench.com/news/anthropic-gives-paid-claude-users-autonomous-browser-actions-2026-08-26