Anthropic's Claude gains Mac control, escalates AI agent race

Anthropic has launched a significant update for its Claude AI, enabling it to directly control a user's Mac. This feature allows Claude to perform actions like clicking buttons, opening applications, and typing, transforming it into a digital operator rather than just a conversational assistant. Available as a research preview for paying subscribers on macOS, this move intensifies the competition among AI companies like OpenAI and Google to develop agents that can actively work within existing software. The system prioritizes direct integrations but can fall back to screen-level interaction, which, while flexible, is slower and more prone to errors.