Anthropic rolls out Cowork preview to let Claude automate complex tasks

Anthropic is launching Cowork for its Claude chatbot as a research preview, a feature that gives the model more agency to automate routine but time-consuming tasks. Cowork is built atop Claude Code and is designed to execute complex functions with minimal human prompting while keeping users updated on the steps it takes.

The company said users can hand over raw materials, grant Claude access to specific folders on their computer, and modify the feature to use connectors, skills, and Google Chrome. Anthropic warned that Cowork carries security risks. The company said Cowork will ask users for confirmation "before taking any significant actions," but cautioned that ambiguous instructions could lead to destructive outcomes — noting that "Claude can take potentially destructive actions (such as deleting local files) if it's instructed to." Anthropic also flagged vulnerability to prompt injection, saying it has built "sophisticated defenses against prompt injections" but that the problem remains "still an active area of development"; OpenAI has written that prompt injections will likely remain an unsolvable problem for AI agents.

The debut of Cowork comes as Anthropic has sought to broaden Claude's appeal beyond engineers and businesses. In September the company announced it had raised $14 billion, bringing its valuation to $183 billion, and The Wall Street Journal reported the company could be valued at $350 billion after a new funding round.

Key Topics

Tech, Anthropic, Claude, Cowork, Claude Code, Prompt Injection