Claude Code & Claude Cowork Get Major Update, Directly Challenging OpenClaw

Anthropic released major updates to Claude Code and Claude Cowork, adding direct computer‑control capabilities that let the AI move from merely giving advice to actually executing tasks, positioning it as a direct competitor to OpenClaw.

The new functionality lets the model open local files, launch browsers and IDEs, control mouse and keyboard, click buttons, and scroll pages without pre‑built tool interfaces. Before acting, it requests user permission and incorporates multi‑layer security, anomaly monitoring, access authorization, and an interrupt mechanism; the feature is currently in research preview for macOS Claude Pro and Max users.

At the same time, the Dispatch feature removes device boundaries, supporting seamless conversation across phone and computer. Combined with computer control, users can remotely command the AI to check email, summarize data, modify code, submit pull requests, and track 3D‑printing progress, enabling true cross‑device automated workflows.

Users praise Claude’s intuitive interaction and its ability to infer vague intents, allowing it to organize files, perform SEO optimization, and complete tasks without explicit step‑by‑step instructions. Industry observers view this as a shift from a chat assistant to a “digital operator,” fundamentally changing human‑computer interaction.

On the same day, OpenAI‑owned OpenClaw announced a comprehensive architecture overhaul: it rebuilt its plugin system, retired the old API, launched an official SDK and the ClawHub plugin store for safety, upgraded models to M 2.7 and GPT‑5.4, improved Windows security, and enabled plugin interoperability with Claude, Codex, and Cursor, moving from a closed framework to an open ecosystem.

These developments crystallize two AI‑agent strategies: Claude follows a product‑focused minimal path emphasizing instant‑use, safety, and human‑like operation to lower the barrier for ordinary users; OpenClaw pursues a systematic open path that emphasizes a plugin ecosystem, multi‑platform compatibility, and long‑running tasks, aiming to become an AI operating system.

The simultaneous upgrades sparked heated industry debate. Some claim Claude’s ease of use “crushes” OpenClaw, while others see long‑term value in OpenClaw’s open‑ecosystem approach. Security and stability remain core concerns—mis‑operation risks, permission control, and data safety—and the macOS‑only limitation leaves many Windows users disappointed.