Anthropic has announced the release of an updated version of its Claude AI model alongside a beta for computer control. The new update allows developers to use Claude AI for tasks similar to those performed by a person at a computer.

New Computer Control

The computer control beta unveiled by Anthropic allows developers, via API, to enable Claude to guide the mouse cursor, click on interface elements, and insert text into a user's computing environment. A demonstration video showcases Claude's ability to navigate between windows, open new tabs for information search, and transfer data from one window to another.

Privacy and Security

Anthropic addresses privacy and security concerns in a blog post, noting that computer use may provide a new vector for threats such as spam, misinformation, or fraud. The company states it is taking a proactive approach to ensuring safe deployment and can identify improper use scenarios.

Beta Expectations

Although computer control is currently available only to developers via the Claude API, Anthropic is releasing it early to gather feedback. The company anticipates the feature will improve rapidly over time.

Anthropic's new beta provides developers the opportunity to experiment with deeper AI integration in computer systems. It opens new prospects for automation and increased work efficiency within computing environments.