Cyber Risks · The Hacker News

Securing Gen AI in the Browser: Policy, Isolation, and Data Controls That Actually Work

Summary: The article describes how employees use browser-based GenAI tools and agentic browsers like ChatGPT Atlas to draft emails, summarize documents, write code and analyze data, often by pasting sensitive information into prompts or uploading files, a pattern traditional security controls were not built to inspect. It lays out a three-part approach - enforceable policy categorizing sanctioned and disallowed tools, isolation through dedicated browser profiles and per-site controls, and data controls that inspect copy/paste, drag-and-drop and file uploads at the browser edge - plus default-deny handling of AI-powered extensions and SSO tied to corporate identities.

Key Takeaway: AI-powered browser extensions often require permissions to read and modify page content, keystrokes and clipboard data, which without oversight makes them an exfiltration channel for sensitive information. The article's own rollout path gives organizations roughly 30 days to move from ad-hoc GenAI use to a formal browser policy with alerts wired into SOC workflows.

Read the original at The Hacker News ↗

Topics
AI Security
Related