The Age of the All-Access AI Agent Is Here

The report examines the growing privacy and cybersecurity risks posed by generative AI agents and assistants that demand extensive OS-level permissions and access to personal and enterprise data (emails, calendars, files, Slack, code, and cloud storage) to function. It highlights the shift from simple chatbots to more autonomous, action-taking systems, notes their current unreliability, and points to examples like Microsoft's Recall (frequent desktop screenshots) and Tinder's photo scanning, while experts warn that users have limited ability to verify or control how their data is handled.