Anthropic has introduced a new capability that allows its AI assistant Claude to directly operate a user’s computer and perform tasks.
Claude can move the cursor, click buttons, type, and navigate apps, just like a human interacting with a computer interface.
The AI can open software, browse websites, fill forms, and execute workflows, going beyond simple chat-based assistance.
This feature marks a major leap toward agentic AI, where systems can plan and complete real-world tasks independently.
Users must grant access before Claude takes actions, ensuring control and safety while using the feature.
The “computer use” feature is currently in early testing (research preview) and is expected to improve with feedback and wider rollout.