Claude in Chrome: Anthropic Opens Scheduled Browser Tasks to All Paying Subscribers

Anthropic has opened the Claude in Chrome beta to all paying subscribers, including Pro, Team, and Enterprise plan holders. The extension gives Claude the ability to operate directly within the browser and execute complex multi-step tasks.

The most notable new capability is scheduled browser tasks. Users can set up recurring workflows that run automatically on a defined schedule, without requiring the user to be present. Claude can navigate websites, click buttons, fill forms, extract data, and execute multi-step workflows.

How it works:

Claude in Chrome operates by visually "seeing" the screen as a human user would, rather than reading the underlying code. This enables it to handle websites that do not expose public APIs.

Users can choose between two modes: manual control, where Claude waits for instructions, and Follow a plan mode, where Claude executes a predefined workflow autonomously within defined boundaries.

Practical use cases:

Email management automation, customer support negotiation, competitor monitoring, price data extraction, and regular report generation are among the most frequently cited use cases from early adopters.

Security considerations:

The integration also introduces new risks. Prompt injection attacks, where malicious websites attempt to trick the AI agent into performing unintended actions, are a real concern. Anthropic has implemented countermeasures, but the field is evolving rapidly.

For CIOs:

This represents a paradigm shift in automation. Traditional RPA required carefully configured robots for each system. Claude in Chrome can theoretically be authorized to work across all web-based systems, including legacy solutions without an API. That opens opportunities, but demands clear security policies and access control frameworks.