Anthropic has launched the Claude 3.5 Sonnet model, capable of controlling desktop applications through a new 'Computer Use' API now in open beta. The model can automate tasks by simulating keystrokes and mouse movements, offering potential for developers to enhance productivity. While the AI shows promise, it struggles with basic actions and has a mixed success rate in completing tasks. Despite concerns about misuse, Anthropic aims to learn from real-world applications to improve safety. The company also announced an upcoming version of its efficient Claude 3.5 Haiku model, expected to enhance user interactions.