30s Summary

AI company Anthropic has launched a developer’s beta of its Claude AI model, enabling independent control of a computer’s mouse, buttons, and typing functions. The new capabilities mean Claude can manage multiple windows, conduct online searches, and move data autonomously. Currently only available via Claude’s API, Anthropic released the beta to gain developer feedback. Although the feature might expose users to spam, misinformation, or fraud, the company said it has prioritised the safe deployment of the new feature.

Full Article

Anthropic, a company specializing in artificial intelligence, recently introduced a new version of their Claude AI model. They also launched a developer’s beta for “computer use” on October 22.

This computer use beta provides developers a chance to equip Claude with the ability to independently control their computer’s mouse, click on various buttons and fields, and autonomously type text into their computer.

A video demo of Claude’s computer use shows off the AI’s skills, such as managing multiple windows, opening new tabs to search the internet, and moving data from one window to another.

Artificial Intelligence models like Claude or OpenAI’s ChatGPT usually interact with users through a dedicated window. While some models can perform limited web searches within their interface, they mainly allow users to upload documents that the AI can parse.

But with Anthropic’s new computer use beta, the company is hoping that developers will be able to automate any task that a person could do at a computer. According to the company’s blog post, “Developers can direct Claude to use computers just like a human does — look at a screen, move a cursor, click buttons, and type out text. Claude 3.5 Sonnet is the first AI of its kind to offer computer use in public beta. Remember, it’s still experimental and may have some kinks to work out.”

Right now, the computer use beta is only available to developers through the Claude API. Anthropic decided to release it early to gather feedback from developers and the company expects the feature to become better over time.

The company also addressed privacy and security concerns in their blog post. They acknowledged that computer use could potentially expose users to issues like spam, misinformation, or fraud. To prevent this, the company is prioritizing the safe deployment of this new feature and can identify if computer use is being misused.

Source: Cointelegraph