Who is involved
Before the recent update, users of Claude AI had a limited interaction with their computers, primarily relying on traditional input methods. The initial expectations revolved around Claude serving as an assistant, providing suggestions and automating simple tasks. The launch of Anthropic’s earlier ‘computer use’ API tool in 2024 was a significant step, but it did not allow for direct control over computer systems.
On March 24, 2026, a decisive moment arrived with the announcement of the auto mode for Claude Code, which was further enhanced in the Claude AI update on March 25, 2026. This update allows Claude to take direct control of users’ computers through the Cowork and Claude Code tools. Users can now experience Claude moving the mouse, clicking, typing, and interacting with various UI elements, marking a substantial shift in user experience.
The immediate effects of this update have been profound. Users describe their interactions with Claude as ‘hypnotic’ or ‘magic,’ particularly in the realm of knowledge work. The AI’s ability to handle repetitive or multi-step tasks across applications has transformed workflows, enabling users to focus on more complex aspects of their work. Claude can now autonomously make decisions regarding file writes and bash commands, applying internal safety classifiers to ensure actions are appropriate.
Experts have noted that this update represents a significant advancement toward practical AI agents capable of operating computers autonomously. Matthew Schwartz, a researcher involved in testing Claude, stated, “AI is not doing end-to-end science yet. But this project proves that I could create a set of prompts that can get Claude to do frontier science. This wasn’t true three months ago.” This perspective highlights the rapid evolution of AI capabilities and their potential applications in complex fields like theoretical physics.
Claude Opus 4.5 was utilized in a theoretical physics research workflow, demonstrating its ability to perform calculations, generate code, conduct numerical checks, and assist in manuscript drafting. The final paper produced from this workflow was deemed a valuable contribution to quantum field theory, showcasing the practical implications of the AI’s capabilities.
However, the update is not without its caveats. It is currently available only to Claude Pro and Max subscribers on macOS, with Windows support expected to follow, although the exact timeline remains unclear. Additionally, the feature is labeled as an early research preview, meaning it may contain bugs and rate limits, which could affect user experience.
As this technology continues to evolve, there are uncertainties regarding the long-term implications of granting AI full control over computers. Questions about job displacement and accountability arise as users become increasingly reliant on AI for critical tasks. The balance between leveraging AI’s capabilities and maintaining human oversight will be a crucial consideration for future developments.
In summary, the Claude AI update signifies a transformative moment in the relationship between humans and technology, pushing the boundaries of what is possible in computer interaction. As users adapt to these new capabilities, the landscape of productivity and automation is poised for significant change.