Reported about 1 month ago
Anthropic has introduced a new AI tool capable of interpreting a user's computer screen and performing tasks on their behalf, advancing the concept of AI agents beyond typical chatbots. This new capability, termed 'computer use,' allows the AI to browse the web, click buttons, and even send calendar invites by comprehending the user's screen activity. While the tool aims to increase productivity, it raises security concerns regarding user privacy and potential errors in execution, prompting Anthropic to implement safeguards. The beta version is initially available to developers using the company's Claude technology.
Source: YAHOO