Claude takes another important step
Anthropic has announced a notable feature for AI assistant Claude with the ability to directly control the user's computer to perform specific tasks.
This is considered an important step forward, moving AI from the role of answering questions to actively operating like a true digital assistant.
According to Anthropic, users only need to register for appropriate service packages to allow Claude to access the computer to perform tasks such as searching and sending files, opening applications or operating on the browser.
This AI can interact with the system by scrolling, clicking or typing keys, similar to how humans use the keyboard and mouse.
Anthropic's move takes place in the context of AI platforms shifting to autonomy capabilities.
Previously, the OpenClaw open source ecosystem created a wave of tools that allowed AI to perform tasks automatically on computers.
Recently, Nvidia also joined the game with NemoClaw, helping to deploy these tools more easily with security layers.
Claude is designed to connect to many popular applications such as Google Calendar or Slack.
In case tools are not available, AI can still manually operate on the interface, expanding the ability to handle tasks more flexibly.
Anthropic also integrates features with Dispatch, allowing users to assign tasks remotely via phone, such as checking morning emails or starting work sessions.
This combination helps Claude handle work even when users do not directly use the computer.
Risk-added utilities
Despite bringing many conveniences, allowing AI to access and control personal computers also raises security concerns.
Experts warn that AI can perform large-scale actions in a short time, even beyond the user's expectations.
The risk is even greater if the system is exploited by hackers, leading to the illegal use of personal data or computer resources.
In particular, forms of attacks such as instantly inserting malware can take advantage of AI to spread quickly.
Anthropic said it has deployed many layers of protection, including mechanisms to detect vulnerabilities and prevent attacks.
However, the company also admitted that this feature is still in the testing phase and there may be errors.
Limited testing, caution when using
Currently, Claude's computer control feature is only provided as a research preview for Claude Pro and Claude Max users, and is limited to devices running macOS.
Anthropic recommends that users should not use this feature with applications containing sensitive data.
Some functions related to important information have even been disabled by default to minimize risks.
Although there are still many limitations, this step shows that the clear trend is AI gradually shifting from supporting tools to entities capable of independent action.
If completed, systems like Claude could change the way people interact with computers, but go hand in hand with higher requirements for control and safety.
