Asger Risom
Anthropic has unveiled a new capability for its Claude AI model, allowing it to use a user’s computer to perform tasks automatically as the race to build autonomous AI agents accelerates.
Artificial intelligence is taking a step closer to acting on behalf of users, not just responding to them.
A new update from Anthropic signals a shift toward more autonomous systems capable of handling everyday digital tasks.
According to CNBC, cited by News.ro, Anthropic has introduced a feature allowing its AI model Claude to directly use a person’s computer. Users can assign tasks remotely, including from their phone, and the system carries them out automatically.
The company said Claude can open applications, browse the web and complete actions such as editing files or filling in spreadsheets.
In a demonstration shared by Anthropic, a user running late asks the AI to export a presentation as a PDF and attach it to a meeting invite. The system completes the process without further input.
Rise Of Agents
The update reflects a broader push across the tech industry to develop so-called AI agents, designed to operate independently around the clock.
These tools gained attention with the launch of OpenClaw, an app that connects to AI models and executes commands through platforms like WhatsApp and Telegram.
Like Claude’s new capability, OpenClaw runs locally on devices, giving it access to user files.
Nvidia CEO Jensen Huang recently told CNBC that OpenClaw is “definitely the next ChatGPT,” as companies compete to build similar systems.
Industry Race
The competition is intensifying. Nvidia has introduced NemoClaw for enterprise use, while OpenAI recently hired OpenClaw creator Peter Steinberger to accelerate its own agent development.
Anthropic’s update positions Claude within this fast-moving space, where companies are racing to define the next major interface for AI.
The goal is to move beyond chat-based assistants toward systems that can act directly on a user’s behalf.
Risks Remain
Anthropic cautioned that the feature is still developing and not as advanced as some of Claude’s other capabilities.
“Claude can make mistakes and while we continue to improve security measures, threats are constantly evolving,” the company said.
It added that safeguards are in place, including requiring user permission before accessing new applications.
The company also highlighted its “Dispatch” feature, which enables ongoing interaction with Claude across devices while assigning tasks continuously.
Sources: CNBC, News.ro
