Claude AI Update: Code and Cowork Tools Can Now Operate Your Computer Directly

AI company Anthropic announced significant updates to its Claude Code and Claude Cowork tools. These AI assistants can now utilize users' computers to directly complete tasks, marking a step from passive responses to active operations. This update aims to enhance work efficiency by allowing the Claude AI chatbot to simulate user actions in computer applications and systems after authorization.

According to the announcement, with this feature enabled, Claude will prioritize connecting to supported services like Google Workspace or Slack to execute tasks. If a dedicated connector for a specific service is unavailable, it will still attempt to fulfill instructions through underlying computer access. Anthropic emphasizes that Claude requests explicit user permission before taking action and advises against handling sensitive information with this feature for security. Currently, this computer control function is only available to Claude Pro and Claude Max subscribers on macOS and remains in a research preview phase, with performance being refined based on user feedback.

This development reflects the trend of AI assistants evolving into more integrated, proactive work partners. Claude Cowork, launched in January for a broader non-professional user base, gaining computer control capabilities further blurs human-computer interaction boundaries. However, balancing convenience with privacy and security will be key for future adoption of such features. As technology iterates, the supported platforms and task ranges are expected to expand gradually.