{"id":208546,"date":"2026-03-25T12:16:00","date_gmt":"2026-03-25T11:16:00","guid":{"rendered":"https:\/\/liora.io\/en\/anthropic-claude-ai-autonomous-computer-control"},"modified":"2026-03-25T12:16:00","modified_gmt":"2026-03-25T11:16:00","slug":"anthropic-claude-ai-autonomous-computer-control","status":"publish","type":"post","link":"https:\/\/liora.io\/en\/anthropic-claude-ai-autonomous-computer-control","title":{"rendered":"Anthropic’s Claude AI sparks autonomous computer control shift"},"content":{"rendered":"

\nAnthropic launched a new feature Tuesday that lets its AI assistants Claude Code and Claude Cowork directly control users’ computers, including mouse movements, keyboard inputs, and screen navigation. The autonomous computer control capability, available now as a research preview for Claude Pro and Max subscribers on macOS, requires explicit user permission before taking any action.\n<\/strong><\/p>\n

The breakthrough feature transforms Claude Code<\/b> and Claude Cowork<\/b> from traditional chatbots into active digital assistants that can execute complex tasks independently. According to The Verge, users can now delegate assignments from their phones, and the Claude desktop app will complete them autonomously on their Mac computers, even when users are away from their devices.<\/p>\n\n

How It Works<\/h2>\n\n

The technology operates by allowing Claude to visually interpret the computer screen<\/a> and interact with it just as a human would, Anthropic explained in its announcement. The AI can move the cursor, click buttons, input text, open files, navigate web browsers, and interact with various applications on macOS.<\/p>

What sets this apart is the “no setup required”<\/b> approach. Rather than relying on pre-built API integrations for every application, Claude operates at the user interface level, according to The Verge. While the system prioritizes direct connectors for supported services like Slack<\/b> and Google Workspace<\/b>, it defaults to direct UI control when connectors are unavailable, making it broadly applicable across most software without extensive configuration.<\/p>

For developers specifically, Claude Code<\/b> can operate developer tools directly, potentially streamlining coding and development workflows, The Verge reported.<\/p>\n\n

Safety and Access Controls<\/h2>
\"Computer<\/figure>\n\n

Anthropic has implemented strict permission controls for the feature. The system will “always ask for your explicit permission” before exploring, scrolling, or clicking to complete tasks, according to the company’s announcement. This permission-based approach addresses potential security concerns about autonomous AI control.<\/p>

The feature is currently in research preview<\/b> phase, allowing Anthropic to gather performance data and identify areas for improvement. Access remains limited to Claude Pro<\/b> and Claude Max<\/b> subscribers using the desktop app on macOS.<\/p>\n\n

Industry Implications<\/h2>\n\n

This development marks a significant shift in AI assistant capabilities, moving beyond conversational interfaces to direct computer manipulation<\/a>. The technology could reshape how users interact with their devices, potentially automating routine digital tasks across multiple applications without requiring technical expertise or complex setup procedures.<\/p>

The cautious rollout as a research preview suggests Anthropic is taking a measured approach to deploying this powerful capability, likely monitoring for potential risks and gathering user feedback before broader availability. Similar ambitions were seen with OpenAI’s autonomous computer-controlling agents<\/a>, underlining how the race to agentic AI is intensifying across the industry.<\/p>\n

\n

\n Sources\n <\/h3>\n
    \n
  • anthropic.com\/news<\/li>
  • theverge.com<\/li>\n <\/ul>\n<\/div>","protected":false},"excerpt":{"rendered":"

    Anthropic launched a new feature Tuesday that lets its AI assistants Claude Code and Claude Cowork directly control users’ computers, including mouse movements, keyboard inputs, and screen navigation. The autonomous computer control capability, available now as a research preview for Claude Pro and Max subscribers on macOS, requires explicit user permission before taking any action.<\/p>\n","protected":false},"author":87,"featured_media":208545,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"editor_notices":[],"footnotes":""},"categories":[2417],"class_list":["post-208546","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news"],"acf":[],"_links":{"self":[{"href":"https:\/\/liora.io\/en\/wp-json\/wp\/v2\/posts\/208546","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/liora.io\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/liora.io\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/liora.io\/en\/wp-json\/wp\/v2\/users\/87"}],"replies":[{"embeddable":true,"href":"https:\/\/liora.io\/en\/wp-json\/wp\/v2\/comments?post=208546"}],"version-history":[{"count":0,"href":"https:\/\/liora.io\/en\/wp-json\/wp\/v2\/posts\/208546\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/liora.io\/en\/wp-json\/wp\/v2\/media\/208545"}],"wp:attachment":[{"href":"https:\/\/liora.io\/en\/wp-json\/wp\/v2\/media?parent=208546"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/liora.io\/en\/wp-json\/wp\/v2\/categories?post=208546"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}